25 results about "Surrogate data" patented technology

The invention discloses a deep learning-based road and bridge construction measurement data processing method and system, and the method comprises the steps: extracting spatial topological features in point cloud data through a GCN graph convolutional network, extracting texture and structural features in image data through improved ResNet-50, extracting time sequence features in sensor data based on an LSTM long-short-term memory network, and carrying out the deep learning-based road and bridge construction measurement data processing. Introducing a multi-head self-attention mechanism to dynamically calculate weights of different modal features; constructing a VAE variational auto-encoder based on the fusion feature vector, judging an abnormal point through a reconstruction error, and performing secondary verification on the abnormal point in combination with a bridge structure mechanical model; and adopting a GAN generative adversarial network to generate alternative data of the abnormal point detection result, and selecting an optimal correction scheme through Bayesian optimization. And the data processing efficiency and the abnormal point detection accuracy are improved.

The invention relates to a method and a system for detecting nonlinear modulation of tide to sea waves, processing equipment and a storage medium. The method comprises the following steps: acquiring an observation data set and preprocessing the observation data set; carrying out holographic marginal spectrum analysis on the preprocessed observation data set, and determining a holographic marginal spectrum of each piece of observation data in the observation data set; generating a plurality of groups of alternative data for the preprocessed observation data set; holographic marginal spectrum analysis is conducted on the generated multiple sets of replacement data, corresponding holographic marginal spectrums are calculated respectively, the # imgabs0 # percentile of the holographic marginal spectrum values of the multiple sets of replacement data serves as the # imgabs1 #%-confidence level of the holographic marginal spectrum values of the observation data, and # imgabs2 # is the preset significance level; and for the same modulation frequency, if the holographic marginal spectrum value of the observation data is smaller than # imgabs3 #%-confidence level, the modulation of the modulation frequency on the observation data is statistically incredible, otherwise, the modulation of the modulation frequency on the observation data is statistically credible, and the method can be widely applied to the field of signal processing.

A method of dynamically optimizing decision tree inference is provided. The method, which is performed at the computerized system, repeatedly executes one or more decision trees for inference purposes and repeatedly performs an optimization procedure according to two-phase cycles. Each cycle includes two alternating phases, i.e., a first phase followed by a second phase. The decision trees are executed based on a reference data structure, whereby attributes of nodes of the decision trees are repeatedly accessed from the reference data structure during the first phase of each of the cycles. First, the accessed attributes are monitored during the first phase of each cycle, which leads to update statistical characteristics of the nodes. Second, a substitute data structure is configured during the second phase of each cycle based on the updated statistical characteristics. Third, the reference data structure is updated in accordance with the substitute data structure.

The invention discloses a historical data-based substitute training machine learning early warning method, which comprises the following steps of: on the basis of a small amount of historical data of a real world system, generating substitute data of a plurality of time sequences with the same statistical attribute as the real historical data; and then training a machine learning classifier based on the generated substitute data to identify different dynamic characteristics between a pre-critical-transition stage and a critical stage, thereby detecting an early warning signal of a critical state or phase change. The detection method is applied to experience and experimental data sets of geology, climatic, sociology and cardiology, the effectiveness of the detection method is verified, and the sensitivity and the specificity of the detection method are higher than those of two widely used general early warning signals. Therefore, the method is not limited by partial bifurcation hypothesis in the conventional method; the method based on the specific system can improve the early warning signal.

The present invention discloses a method, a system and a storage medium for early fault detection of a hydroelectric generating unit, belonging to the technical field of mechanical fault diagnosis methods. The method includes: collecting vibration data of a water turbine and using the vibration data as source data; performing blind source decomposition on the source data based on a binary decomposition method to decompose it into multiple source signals; respectively performing surrogate data processing based on the fast Fourier transform on the decomposed multiple source signals to obtain surrogate data; and performing non-linear structural damage determination and fault propagation path drawing based on the surrogate data. The present invention is beneficial to the discovery and maintenance of early faults, thereby reducing the maintenance cost of the hydroelectric generating unit.

The application provides an RTK positioning enhancement method and system, which is used in the field of satellite navigation, and the method comprises the following steps: according to the reference station pseudorange observation data, phase observation value, real-time precise ephemeris, real-time clock difference and OSB product, the atmospheric delay parameter is solved, and the real ionospheric delay and the tropospheric delay in the atmospheric delay parameter are extracted; the dynamic error accuracy of each error source is estimated through the time series analysis method and the space interpolation method; the comprehensive error is calculated, the mathematical relationship model of the comprehensive error and the equivalent SNR is constructed, the equivalent SNR value is updated according to the dynamic error accuracy of each error source, the equivalent SNR is mapped into the SNR field in the RTK data format; the RTK data is encoded, the original measurement SNR value in the RTK data is replaced by the equivalent SNR, and the RTK data is broadcast to the user end. The scheme can realize the precision information transmission without changing the traditional RTK data format, guarantee the positioning reliability under the complex atmospheric condition, and improve the RTK positioning precision.

This invention belongs to the field of network communication security technology, specifically relating to a network communication security protection method and system based on plaintext data. The method involves acquiring and marking data, identifying abnormal communication nodes based on abnormal data, extracting alternative data from normal data affected by communication interruptions, classifying it into delayed data and data awaiting confirmation, determining whether it constitutes suspected attack event data, constructing a classification standard based on impact degree scoring to rank the suspected attack event data, extracting its original content as verification information for repeated review, and outputting alarm information or corrected normal data based on the verification results. This invention achieves high accuracy in attack event identification, high efficiency and targeted security protection, and high reliability and system robustness in the final judgment through layered filtering and classification, combined with a dual-priority processing mechanism of initial risk screening and impact degree ranking, and a closed-loop confirmation and correction process.

In a machine learning apparatus that learns data of a novel class with a smaller number of samples than data of a base class by continual learning, a feature extraction unit is pre-trained using the data of the base class. The feature extraction unit receives an input of the data of the novel class to output a feature vector of the data of the novel class. A weight calculation unit calculates a classification weight of the novel class based on the feature vector. A graph model receives an input of the classification weight of the novel class and classification weights of all classes previously learned to output reconstructed classification weights. The graph model is trained by pseudo continual learning using alternative data of the base class to learn a dependency between the base class and the novel class by meta learning.

Apparatus for ensuring data accuracy and related methods include a processor and a memory connected to the processor, wherein the memory contains instructions configuring the processor to receive a data structure with a plurality of data attributes, validate the data structure, using a validation machine-learning model, by comparing each data attribute of the plurality of data attributes against at least a validation metric, identify an error by identifying at least an erroneous data attribute as a function of the validation, display, using a user interface, an annotation as a function of the error, wherein the annotation comprises an indication of rationale, generate at least a substitute data attribute for replacing the at least an erroneous data attribute, receive supplemental input in response to the at least a substitute data attribute, and update the data structure by resolving the error as a function of the supplemental input.

The embodiment of the invention provides a safety assessment method and device for joint training. In the process that a first party carries out first joint training on a local first model and a second model of a second party, a pseudo-second model used for simulating the second model is locally trained based on the first model. Then, inputting the first simulated user data into the first model, inputting the second simulated user data into the pseudo second model for joint prediction, and determining first gradient information of the first model based on a joint prediction result and the pseudo second model; the first party inputs the first simulation user data and replacement data used for replacing the second simulation user data into a first reconstruction model to obtain initial reconstruction data for the second simulation user data; and fusing the potential representation of the initial reconstruction data and the potential representation of the first gradient information through a second reconstruction model, and determining predicted reconstruction data based on the fused representation. And the first party evaluates the security of the first joint training according to the similarity between the predicted reconstruction data and the second simulation user data. Privacy protection needs to be carried out on privacy data in the evaluation process.

The invention relates to a method for processing at least one series of temporal data, such as daily data, representative of one or more physical quantity(ies) influenced by seasonal effects and implemented in a given industrial process, the method comprising: - a first modified STL internal loop composed of seven steps focused on the extraction of a first version of the trend and seasonal components, and - a second HTL (Holiday Trend Decomposition Loop) internal loop composed of six steps and which disentangles the outliers of the holiday estimates and leads to a systemic construction of series corrected for seasonal variations and holidays; - the generation of a trend indicator adapted to the given industrial process, the indicator being accompanied by a predetermined threshold beyond which an action is programmed to be undertaken in said industrial process Figure for the abstract: Fig. 1

The application discloses a kind of hierarchical system data migration method, device, equipment and storage medium, it is applied to hierarchical system total center in the field of data migration, comprising: obtaining the cascade registration request sent by target subcenter is cascaded, and according to the data migration task of surrogate migration data is generated;Get migration task execution instruction, and the corresponding surrogate migration data and the agent information of several to-be-migrated collection agents are issued to target subcenter;Data receiving information returned by target subcenter is acquired, cascade instruction is issued to to-be-migrated collection agent to realize the cascade of to-be-migrated collection agent and target subcenter, and target information that data migration succeeds is generated, and the end of surrogate data migration task is marked according to target information.Through the data migration between nodes and node cascade, the information and the right of original node are effectively and safely migrated to new node, while the function of original node is not interrupted, new node can obtain the management right of the data and assets of original node.

The invention relates to a method for predicting a subsea pipeline vibration and local scouring coupling process, and belongs to the technical field of ocean monitoring. The method comprises the following steps: training a PINN model fusing structure vibration-flow field change-scouring development physical information, and finally, in the training process, through minimizing a neural network loss function, predicting the local scouring coupling process of a subsea pipeline. And training a PINN model fusing structural vibration-flow field change-scouring development physical information, and carrying out numerical simulation on the vortex-induced vibration and local scouring long-duration process of the elastic submarine pipeline. According to the method, physically-guided data learning is used for replacing data-driven physical solution, so that the physical reliability of traditional numerical simulation is reserved, and the high efficiency and generalization of a data model are achieved. For the engineering problems of high risk, difficult monitoring and strong coupling such as subsea pipelines, a rapid tool can be provided for multi-scheme comparison and selection in the pipeline design stage, and a feasible technical path can be provided for vibration-scouring risk real-time early warning in the operation stage.

The application relates to the technical field of data processing, and discloses an arteriovenous fistula remote monitoring and warning system which comprises the following steps: collecting monitoring data of multiple dimensions of an arteriovenous fistula; performing graph clustering on all-dimensional monitoring data at each moment to obtain a category division result at each moment, obtaining a change point and a change abnormality rate according to the category division result, and obtaining an abnormal time period, abnormal data and an abnormal dimension sequence; performing EMD decomposition on each abnormal dimension sequence to obtain a plurality of IMF components, obtaining a cycle length of each IMF component, obtaining replacement data and a conversion vector of each segment of abnormal data according to the cycle length, and compressing the replacement data and the monitoring data to obtain compressed monitoring data; and completing the arteriovenous fistula remote monitoring and warning according to the compressed monitoring data and the conversion vector. The application aims to solve the problem that abnormal data destroys the regularity of monitoring data and affects the compression efficiency and transmission efficiency.

Systems, computer program products, and methods are described herein for implementing dynamic data obfuscation using pattern recognition techniques. The present invention is configured to electronically receive one or more data artifacts; electronically receive one or more masked data artifacts; initiate one or more machine learning algorithms on the one or more data artifacts and the one or more masked data artifacts; determine, using the one or more machine learning algorithms, a first set of patterns associated with the one or more data artifacts and a second set of patterns associated with the one or more masked data artifacts; determine a similarity index between the first set of patterns and the second set of patterns; and compare the similarity index with a predetermined threshold; determine one or more alternate data obfuscation algorithms; and implement the one or more alternate data obfuscation algorithms on the one or more data artifacts.

The invention belongs to the technical field of network communication security, and particularly relates to a network communication security protection method and system based on plaintext data. The method comprises the following steps: acquiring data and marking the data, identifying an abnormal communication node according to abnormal data, extracting alternative data from normal data influenced by communication interruption, dividing the alternative data into delay data and to-be-confirmed data, and judging whether the alternative data form suspected attack event data or not; constructing a classification standard based on an influence degree score to arrange the suspected attack event data; and extracting the original content as verification information to repeatedly execute review, and outputting alarm information or corrected normal data according to a verification result. According to the method, through layered filtering and classification, a dual priority processing mechanism combining risk preliminary screening and influence degree sorting, and a closed-loop confirmation and correction process, high accuracy of attack event identification, high efficiency and pertinence of security protection, high reliability of final judgment and system robustness are realized.

A unified method for neural operator training and partial differential equation system solving based on variational principle, medium and product, comprising: sampling from the parameter space of the partial differential equation system to form a data set containing only discrete parameter field, divided into offset set, test set, unlabeled set, the unlabeled set is divided into multiple batches, and a mask tensor of boundary condition is formed; using a neural operator module to predict the node solution of the discrete parameter field sample in the unlabeled set, obtaining a discretized functional as an estimate of the system functional; calculate the gradient of the functional estimate with respect to the node solution, take its norm as the minimization target, obtain the update step of the current node solution, and update the weight of the module. The application unifies the two tasks of solving the partial differential equation system operator and training the neural operator in one framework, introduces a variational operation to construct an unlabeled optimization target to replace the data-driven training error term, saving the time and computing power required to generate a large number of labels for constructing the data-driven error term.

By making it possible to efficiently generate program code without leaking user 6's mesh data to the outside, it is possible to achieve both reliability and convenience of services on the distributed mesh data platform. [Solution] The support device 1 has a control unit 11 which includes a substitute data generation command unit 113 which commands a language model to generate substitute data that has a data structure common to mesh data specified by a user 6, a code generation command unit 114 which commands the language model to generate program code that performs specified data processing on the substitute data, an execution result generation unit 119 which generates the execution result of the program code, a processing target designation unit 118 which designates specified processing target data from a group including substitute data and mesh data as data to be executed by the program code in the execution result generation unit 119, and a mesh data acquisition unit 121 which acquires the specified mesh data.

The invention discloses an IBIS model data processing method and apparatus, a computer device and a storage medium. The method comprises the steps of obtaining a pull-up curve and a pull-down curve of a target IBIS model; judging the effectiveness of the pull-up curve and the pull-down curve; in response to the judgment that only the pull-up curve is a valid curve and the pull-down curve is an invalid curve, generating a replacement pull-down curve based on a preset complementary relationship according to the numerical value of the pull-up curve; in response to the judgment that only the pull-down curve is a valid curve and the pull-up curve is an invalid curve, generating a replacement pull-up curve based on a preset complementary relationship according to the numerical value of the pull-down curve; wherein the effective curve and the generated alternative curve are used for circuit simulation. According to the method, under the condition that a curve is missing, available alternative data is generated through reasonable logic inference and construction, it is ensured that the simulation process can continue to be carried out, and a relatively accurate simulation result meeting physical significance is obtained.

The invention provides an RTK (Real-Time Kinematic) positioning enhancement method and system, which are used in the field of satellite navigation, and the method comprises the steps: resolving atmospheric delay parameters according to pseudo-range observation data of a base station, a phase observation value, a real-time precise ephemeris, a real-time clock error and an OSB (Oriented Standard Board) product, and extracting real ionized layer delay and troposphere delay in the atmospheric delay parameters; estimating the dynamic error precision of each error source through a time sequence analysis method and a spatial interpolation method; calculating a comprehensive error, constructing a mathematical relationship model of the comprehensive error and the equivalent SNR, updating an equivalent SNR value according to the dynamic error precision of each error source, and mapping the equivalent SNR into an SNR field in an RTK data format; and encoding the RTK data, replacing the original measurement SNR value in the RTK data with the equivalent SNR, and broadcasting the RTK data to a user side. According to the scheme, on the premise that a traditional RTK data format is not changed, precision information transmission can be achieved, the positioning reliability under the complex atmospheric condition is guaranteed, and the RTK positioning precision is improved.

The invention relates to the technical field of ground vehicle navigation and positioning, in particular to an integrated navigation method integrating fault recovery and multi-task uncertainty estimation, and solves the problems that when an existing IMU / ODO integrated navigation system seriously distorts ODO data due to slipping, a traditional FDI method only removes and does not supplement the ODO data, the system is forced to be degraded into a pure inertial navigation resolving mode, and the navigation efficiency is low. The invention provides an integrated navigation method integrating fault recovery and multi-task uncertainty estimation, which has slip sensing and intelligent recovery capabilities, organically integrates three core links of'detection, replacement and fusion ', and comprises the following steps: data initialization, detection, replacement and fusion. ODO anomaly detection based on innovation is executed, multi-task pseudo odometer replacement data is generated, according to an ODO anomaly detection result based on innovation, an observation mode is dynamically selected, a measurement noise covariance matrix is constructed, IEKF is adopted as a data fusion center, and measurement updating and time recursion are carried out.

This invention discloses a method and related equipment for handling CEMS carbon emission data failures, belonging to the field of online carbon emission monitoring and data quality assurance. The method includes: constructing a working condition knowledge base containing historical "backtracking window-future sequence" key-value pairs; real-time monitoring and identification of CEMS carbon emission data failures; retrieving the K most similar historical segments from the knowledge base as evidence using the backtracking window before failure as the query key; fusing the backtracking window and the retrieved evidence into a time-series prediction model to generate predicted values ​​for the failure period as replacement data; and simultaneously outputting an interpretability report containing the source, similarity, and contribution of the evidence. The system is an intelligent agent executing this method. This application solves the problems of low accuracy and poor interpretability caused by existing fixed-rule substitution methods by explicitly introducing historical similar working condition evidence to constrain and enhance predictions, significantly improving the accuracy, auditability, and adaptability to complex working conditions of data repair.