45 results about "Mean vector" patented technology

The invention relates to the technical field of environmental parameter measurement, in particular to an indoor environment quality intelligent evaluation method based on multi-dimensional data fusion, which comprises the following steps: step 1, performing time-space synchronous acquisition on multi-source data; 2, a dynamic correlation model is constructed, a parameter coupling matrix is generated based on inter-parameter time-delay cross-correlation analysis, the time-delay cross-correlation analysis is achieved through sequence translation in a sliding time window and correlation coefficient calculation, an environment state class is constructed according to a parameter mean vector, fluctuation intensity and abnormal event marks, and a dynamic correlation model is constructed; each environment state class is bound with an independent parameter contribution degree weight set; step 3, performing double-stage environment quality evaluation; and step 4, adaptive increment optimization: when the deviation between the comprehensive evaluation index and the subjective evaluation exceeds a threshold value, adjusting a parameter contribution degree weight set of an environment state class, and periodically reconstructing a parameter coupling matrix. The prediction capability is improved by considering the time-delay coupling between the parameters; the system has self-learning and optimization capabilities, and the stability of long-term operation of the system is improved.

The invention discloses a converter batching data generation and optimization method, which comprises the following steps: acquiring an original industrial data set in a converter batching process, preprocessing the original industrial data set, inputting preprocessed training data into an encoder, generating a mean vector and a variance vector of potential variable distribution, and optimizing the mean vector and the variance vector; constraining the similarity between the potential variable distribution and the standard Gaussian distribution through KL divergence loss; inputting the potential variable and the random noise vector into a generator to generate a simulation sample, and distinguishing a real sample from the generated sample through a discriminator; a forced discriminator module is introduced, and a discrimination result is converted into an additional loss item; and carrying out data generation on the missing batching parameters, outputting a complete batching data set conforming to process constraints, constructing a cost objective function, a quality objective function and a resource consumption objective function, constructing constraint conditions of chemical element content of a target finished product, optimizing the converter batching data, and generating optimized converter batching data.

The invention discloses an equipment operation safety detection method and system of an energy storage system, and relates to the technical field of energy storage system safety detection.The method comprises the steps that multichannel original data are collected and preprocessed, and multichannel signals are obtained; extracting an instantaneous frequency through adaptive decomposition and Hilbert transformation based on the multi-channel signal, performing modal reconstruction based on the instantaneous frequency, extracting a switch frequency band energy feature, an EMI feature, a DC ripple and a coherent feature, and constructing a fingerprint vector; collecting health state historical data, calculating a mean vector and a covariance matrix, and constructing a health template; based on the health template and the real-time fingerprint vector, analyzing a safety state through dual statistical judgment, and outputting a safety statistical magnitude; and judging abnormity based on the current safety statistics, performing LASSO sparse reconstruction by using a predefined typical fault template, and outputting a fault type. Therefore, efficient and accurate fault diagnosis and automatic alarm are realized.

The invention relates to the field of data processing and anomaly detection, and discloses an enterprise supply chain intelligent traceability management system based on a block chain, and the system comprises a data collection module which is used for collecting state data uploaded to the block chain by each node of the enterprise supply chain, and outputting the state data as a multi-dimensional observation data sequence; the data processing module is used for constructing a standardized data stream in a time sequence form based on the observation data sequence; the parameter modeling module is used for estimating statistical parameters of the current data segment through a sliding time window based on the standardized data stream, and the statistical parameters comprise a mean vector and a covariance matrix; and the anomaly detection module is used for receiving the statistical parameters and mapping the statistical parameters to a Riemannian manifold formed by a statistical parameter space. According to the invention, through combination of sliding window modeling and block chain recording, high-precision identification of abnormal information and credible traceable management of whole-process data are realized.

The invention provides a quality control system based on a multivariate exponential weighted moving average control chart, and belongs to the technical field of quality management and statistical process control. The system comprises a quality data acquisition module, a data preprocessing module, a statistical process control module, an anomaly judgment module and a quality evaluation and improvement module, and is used for monitoring and controlling the production process of a product with a plurality of quality characteristics in real time. The method comprises the following steps of: acquiring quality data of a plurality of quality characteristics in a production process by a system, preprocessing the quality data, and constructing a multivariate quality characteristic data set; estimating a mean vector and a covariance matrix in a controlled state based on the historical quality data, performing weighted updating on the multivariate quality data by adopting a multivariate exponential weighted moving average method, calculating a corresponding MEWMA statistic and generating an MEWMA control chart; and comparing the MEWMA statistical magnitude with a preset control limit to judge whether the production process is in an out-of-control state or not, and outputting early warning information and a corresponding quality evaluation result and improvement suggestion when abnormality is detected. The method can comprehensively analyze related information among multivariate quality characteristics, improves the sensitivity and accuracy of anomaly detection in the production process, and effectively improves the quality control level of the production process in the manufacturing industry.

The invention provides an ISAC system signal detection method based on block diagonal expected propagation, and the method comprises the steps: obtaining a communication signal and a target echo signal transmitted by multi-antenna user equipment through a receiving end, carrying out the whitening of the received signal and noise, and minimizing the interference of a radar to communication; converting the optimized complex signal into a real value system model; initializing a mean vector and a covariance matrix parameter of the expectation propagation algorithm, and setting the number of iterations; calculating diagonal elements of the covariance matrix by adopting a block diagonal Neumann series approximation method; updating a mean vector and adjusting an iteration step length through a dynamic residual factor; and after a preset number of iterations, outputting a final detection signal. According to the method, the complexity is reduced by calculating the diagonal elements of the matrix inversion instead of directly solving the matrix inversion, and a special averaging method is designed for the diagonalization characteristic of the matrix with the similar channel hardening characteristic to optimize the precision.

The invention belongs to the technical field of diabetes diagnosis, and particularly relates to a diabetes prediction model construction method based on ensemble learning and a prediction system. The method comprises the following steps: step 1, carrying out preprocessing and feature screening on a data set; 2, dividing a training set and a test set; 3, searching and extracting positive samples in the training set, randomly generating positive samples obeying normal distribution according to the mean vector and the covariance matrix of the positive samples in the training set, adding the newly generated positive samples into the training set, and continuously repeating until the proportion of the negative samples to the positive samples is balanced; and 4, learning the model by using the processed training set machine. The invention also provides a diabetes prediction system constructed by using the machine learning model. The method can effectively improve the prediction performance of the prediction model through the balance of the sample data and the integrated learning of the single classifier, and has a good application prospect.

The invention provides a multi-channel national secret task scheduling method and system based on dynamic priority, and belongs to the technical field of information security. The method comprises the following steps: determining an initial urgency factor according to remaining time from a national secret task to deadline; determining a dynamic enhancement factor by adopting an adaptive immune optimization enhancement algorithm to correct the initial emergency factor to obtain a final emergency factor; a Bayesian linear regression model corresponding to the Bayesian updating channel is used to generate a new value of a posterior mean vector and a new value of a posterior covariance matrix, a basic suitability factor is determined according to the new value of the posterior mean vector, and the basic suitability factor is corrected according to the new value of the posterior covariance matrix to obtain a final suitability factor; multiplying the dynamic enhancement factor, the final urgency factor and the final suitability factor to obtain the priority of the national secret task; and constructing a global optimal allocation model according to the priorities of the national secret tasks, and solving to obtain a national secret task scheduling strategy. According to the invention, the system resource use efficiency is improved.

The invention discloses a distributed particle filtering algorithm based on Gaussian mixture and Gossip fusion, which relates to the technical field of distributed state estimation and comprises the following steps: acquiring local posteriori distribution by a sensor node through local particle filtering, exchanging sufficient statistics of GMM between the node and a randomly selected neighbor node based on a Gossip protocol, and obtaining a distributed state estimation result. Iterative fusion of the GMM is realized through weighted mixed importance sampling; and finally, optimizing a global state estimation result through a posterior correction step. According to the method, only full statistics such as a mixed coefficient, a mean vector and a covariance matrix of a Gaussian mixture model (GMM) are exchanged among nodes, original particle transmission is replaced, the communication overhead is remarkably reduced, and the communication complexity is linearly increased along with the network scale; meanwhile, an adaptive GMM component adjustment, a regularization weighting EM algorithm and an iteration updating strategy of random weight matrix control are adopted, the calculation process is simplified on the premise that the estimation precision is guaranteed, the single-node calculation burden is reduced, and large-scale sensor network deployment with limited adaptive resources is achieved.

The invention provides an attribute missing target fusion identification method and system based on covariance bidirectional alignment. The system comprises an initialization module, a feature learning module, a soft classification module and an evidence fusion module. The initialization module is used for inputting data and preliminarily filling missing values in test data based on a mean vector of training data; the feature learning module obtains new feature representation of the training data and the filled test data through the feature conversion matrix; the soft classification module is used for training two classifiers to obtain two soft classification results of the test data under the original feature representation and the new feature representation; and the evidence fusion module obtains the optimal weight by minimizing the objective function, and fuses the discounted soft classification result based on the evidence theory to obtain the final category of the test data. According to the method, the alignment process of data distribution is fused into learning of missing values, so that the negative influence of data distribution offset on a downstream target identification task is effectively reduced.

The invention discloses an ion adsorption type rare earth mining area nitrogen pollution source analysis method and device and a storage medium, and can be applied to the cross technical field of environmental geology and mine pollution abatement. After different water bodies are adopted, topographic parameters, physicochemical indexes, nitrogen pollution indexes and isotope data of samples are obtained, principal component score calculation is performed on the to-be-analyzed samples based on a principal component analysis method to obtain principal component scores, and then a first mean vector and a first covariance matrix of the principal component scores are calculated; performing hierarchical clustering analysis on the to-be-analyzed sample according to the principal component score to obtain a clustering sample, and then performing nitrogen pollution source marking on the clustering sample according to the principal component score, the first mean vector and the first covariance matrix, thereby effectively identifying a pollution source and delineating a pollution point; meanwhile, the nitrate nitrogen source in the to-be-analyzed sample is quantified based on the Bayesian isotope mixed model, so that the contribution rate of the nitrogen source can be effectively quantified.

The invention provides a batching method for a flash furnace. The method comprises the following steps: generating a plurality of batching schemes; each batching scheme is input into the flash furnace digital model for analog calculation, and a heat balance parameter corresponding to each batching scheme is obtained; calculating a comprehensive loss value corresponding to each batching scheme so as to update a mean vector and a covariance matrix in a covariance matrix adaptive evolutionary algorithm; judging whether the difference value between the predicted value and the target value of the heat balance parameter of each batching scheme exceeds an allowable error or not; if the batching scheme which does not exceed the allowable error exists, outputting the batching scheme and ending; and if all the batching schemes exceed the allowable error, repeating the above steps according to the updated mean vector and covariance matrix. The batching schemes are generated through an algorithm, simulation calculation is carried out, the heat balance parameter of each batching scheme is predicted, and the batching schemes are evaluated by judging the heat balance parameters, so that the optimal batching scheme is obtained.

The invention relates to the technical field of machine learning data preprocessing, in particular to a machine learning sampling method and system capable of keeping a feature proportional relation, and aims to ensure that the proportional relation between features and target variables is not distorted by dividing category subsets and constraining that sampled feature mean vectors are consistent with original data. And misjudgment of feature importance by the model is avoided. And by verifying the consistency of the covariance matrix, the correlation structure between the features is maintained, and the generalization ability of the model depending on the statistical characteristics is improved. While the problem of sample imbalance is solved, the proportional relation between the features in the original data set and the target variables and the proportional relation between the features are strictly maintained, so that the accuracy and reliability of machine learning model training are improved.

The invention discloses a real-time separation method for static deflection and dynamic deflection of a bridge. The method comprises the following steps: determining an initial value parameter of deflection separation; establishing a correlation model of the temperature signal and the deflection signal through a linear regression mode; predicting the probability distribution of the static deflection and the temperature at the t + 1 moment based on the mean vector and the covariance matrix of the posterior static deflection value and the posterior temperature value at the t moment, and obtaining the mean vector and the covariance matrix of the prior static deflection value and the prior temperature value at the t + 1 moment; calculating a mean vector and a covariance matrix of a posterior static deflection value and a posterior real temperature value at the t + 1 moment by combining a prior prediction result at the t + 1 moment based on the total deflection and the temperature measurement value so as to obtain a corrected static deflection value and a corrected real temperature value at the t + 1 moment, and subtracting the static deflection from the total deflection to obtain a dynamic deflection value; and real-time separation of static deflection and dynamic deflection is realized through iteration. According to the method, the bridge real-time deflection monitoring data is taken as a main object, the information of the temperature change trend is fused in the static deflection separation process through iteration of prior prediction and posterior correction, the separation result basis is sufficient, the algorithm complexity is low, and the engineering practicability is high.

The present invention provides a hyperspectral anomaly detection method based on a dual-branch generative adversarial network, comprising: decomposing a low-rank rough background matrix and a sparse rough anomaly matrix from an original hyperspectral image; inputting the rough background matrix and the rough anomaly matrix into a dual-branch network model, wherein the dual-branch network model outputs a reconstructed background matrix corresponding to the rough background matrix, and the dual-branch network model outputs a reconstructed anomaly matrix corresponding to the rough anomaly matrix; fusing the reconstructed background matrix and the reconstructed anomaly matrix to obtain fused hyperspectral data; calculating a pure background mean vector and a background covariance matrix using the reconstructed background matrix; solving for anomaly response values of each spectral vector in the fused hyperspectral data based on the background mean vector and the background covariance matrix; and performing normalization processing on the anomaly response values of each spectral vector to obtain an anomaly detection result map for the entire hyperspectral image. The present invention enhances the distinguishability between background and anomaly samples and improves anomaly detection accuracy.

The invention relates to the technical field of intelligent ore processing, and discloses a multi-element grade prediction method based on multi-task collaboration and covariance modeling. The method comprises the steps that an ore grade prediction model is constructed, the ore grade prediction model comprises a feature coding layer, a fusion processing layer and an output layer, the feature coding layer is provided with a plurality of encoders for multi-modal data of ore, the fusion processing layer is provided with fusion blocks based on an attention mechanism, and the output layer is constructed based on a multi-layer perception mechanism; the method comprises the following steps: acquiring multi-modal data of to-be-detected ore, inputting the multi-modal data into an ore grade prediction model, respectively acquiring corresponding single-modal features through a plurality of encoders, acquiring fusion features by a fusion processing layer based on the single-modal features, acquiring a mean vector and a covariance matrix by an output layer based on the fusion features, and performing prediction distribution. And analyzing the ore grade data according to the prediction distribution result and regulating and controlling the process flow. The problem that the ore grade cannot be effectively and accurately detected by the existing method for predicting the ore grade by adopting machine learning is solved.

The invention discloses an ion implantation uniformity adjusting method, which comprises the following steps: step 1, establishing a wafer ion implantation amount model, dividing a wafer into n wafer areas and dividing a scanning range of an ion beam into m implantation positions in the model, the model comprising an n * m first matrix, an m * 1 second matrix and an n * 1 third matrix; the first to third matrixes are respectively an injection rate matrix, an injection time matrix and an injection amount matrix, and the third matrix is obtained by multiplying the first matrix by the second matrix. Step 2, performing iterative optimization, including giving values of a second matrix and a third matrix, and calculating a mean vector matrix and a covariance matrix required for enabling the first matrix to conform to normal distribution; and calculating the value of the first matrix through the mean vector matrix and the covariance matrix. And predicting and adjusting the value of the second matrix according to the value of the first matrix, and obtaining the value of the adjusted third matrix according to the value of the first matrix and the adjusted value of the second matrix. The method can reduce the number of iterations.

The embodiment of the invention discloses a fire point detection method based on satellite data and related equipment, and the method comprises the steps: obtaining a plurality of pieces of satellite time series data when there is no fire point, obtaining a potential space vector through a preset encoder, calculating a mean vector and a covariance matrix, and enabling the covariance matrix to describe the discrete degree of data in different directions, the shape of the fire-point-free sample cluster can be described more accurately, and the method is not limited to hypersphere hypothesis any more. For a to-be-detected target point, after target satellite time sequence data and a target potential space vector of the to-be-detected target point are obtained, the submerged space anomaly is obtained in combination with a mean vector and the covariance matrix, and the method for obtaining the submerged space anomaly in combination with the covariance matrix fully considers the actual distribution form of a fire-point-free sample in the submerged space. Therefore, the accurate potential space anomaly degree is obtained, the potential space anomaly degree is input into the preset classifier, the accuracy of fire point detection can be effectively improved, and abnormal point misjudgment caused by false hypothesis is avoided.

The invention relates to the field of computers and communication, and provides a passive Internet of Things positioning method and device, electronic equipment, a medium and a program product, the method comprises the following steps: preprocessing label data collected from passive Internet of Things equipment, the label data comprising data of at least two dimensions; based on the preprocessed label data, respectively estimating a mean vector and a covariance matrix of the label data of the target label; constructing a multivariate Gaussian distribution model corresponding to each target label based on the mean vector and the covariance matrix; performing data enhancement by using a multivariate Gaussian distribution model to generate an enhanced data sample of the target tag; merging the original data sample and the enhanced data sample of the label data to generate an enhanced data sample; constructing a sample set according to the data sample of each label in the passive Internet of Things equipment; and inputting the real-time label data into a positioning model obtained based on sample set training, and outputting a label positioning result. According to the method, effective data enhancement is realized, the sample balance is improved, the accuracy and robustness of a label positioning result are improved, and the positioning precision is improved.

The invention discloses a Bates model parameter estimation method and device and a related product. The method comprises the following steps: determining observation parameters, to-be-estimated parameters and scalar features; dividing intervals for the scalar features, calculating to obtain a cumulative probability corresponding to each interval, and forming an observation statistic vector; wherein the scalar features are extracted from the observation parameters; for the candidate value of any to-be-estimated parameter, simulating n trajectories through a numerical method, and extracting target features of each trajectory; dividing intervals for the target features, calculating to obtain a cumulative probability corresponding to each interval, and forming a first simulation statistic vector; according to each first simulation statistic vector, obtaining a mean vector of the simulation statistic and a covariance matrix of the simulation statistic; and under the condition that the observation statistic vector obeys Gaussian distribution, synthesizing an approximate likelihood function according to the mean vector of the observation statistic vector simulation statistic and the covariance matrix of the simulation statistic, and estimating the model parameters according to the approximate likelihood function.

The invention relates to the field of laser radars, and provides a laser radar point cloud modeling method, system, equipment and medium under a dynamic sea condition, and the method comprises the steps: obtaining unmanned vehicle radar data, unmanned vehicle motion data and water surface environment data of an unmanned vehicle; performing point cloud inter-frame registration and dynamic point elimination to obtain static point cloud data; the method comprises the following steps: performing motion compensation through motion data of an unmanned vehicle in a wake flow area to obtain compensation radar data, performing kernel density estimation to obtain a Poisson wake flow parameter, calculating a Poisson wake flow model parameter through the Poisson wake flow parameter, and obtaining a Poisson term coefficient based on the kernel density estimation; calculating the total number of Gaussian components, calculating the mean vector and Gaussian component weight of the static point cloud data, obtaining a covariance matrix, and constructing a dynamic surface model; and constructing a parallel maximization likelihood function, and performing model parameter estimation through the parallel maximization likelihood function to obtain updated parameters, thereby obtaining target point cloud data and completing modeling of the point cloud data.

The invention discloses a power distribution system collaborative optimization operation method and device, and the method comprises the steps: constructing a generalized energy storage collaborative optimization model of a power distribution system according to the charging and discharging characteristics of an electric vehicle, the charging and discharging characteristics of a fixed energy storage device, and the load characteristics; constructing a context vector learning model containing a mapping relation between context vectors and optimization parameters by using historical operation data of the power distribution system and a generalized energy storage collaborative optimization model of the power distribution system and adopting a Gaussian process regression method; according to the mapping relation, a context vector of the current scene is extracted, initial distribution of parameters is predicted and optimized, and an initial mean vector and a covariance matrix of a CMA-ES algorithm are obtained; and executing a CMA-ES iterative optimization process by adopting the initial mean vector and the covariance matrix, and outputting a scheduling scheme of the electric vehicle and the fixed energy storage device containing the charging and discharging power of each time period. According to the scheme, hot start of the CMA-ES algorithm is realized through context vector learning, the optimization efficiency is improved, and falling into a local optimal solution is avoided.

The application provides a polysilicon production equipment fault diagnosis method and device, a server and a storage medium, and relates to the field of equipment fault diagnosis. The method comprises the following steps: acquiring target equipment collected sample data and first historical sample data within a preset time window; determining a first set composed of equipment parameters and a second set composed of equipment parameters and corresponding upstream and downstream process parameters; further determining the first robust distance of the equipment parameters according to the equipment mean vector and the equipment covariance matrix; determining the second robust distance of the equipment parameters and the corresponding upstream and downstream process parameters according to the overall mean vector and the overall covariance matrix; if the first robust distance is greater than the first robust distance threshold and / or the second robust distance is greater than the second robust distance threshold, the running state is determined to be an abnormal state. The comprehensiveness of the diagnosis and the accuracy of the diagnosis result are ensured, the stability of the benchmark during diagnosis is ensured, and the diagnosis efficiency is improved.

The present invention discloses an abnormal detection method for the geological drilling process based on the spaced augmented Mahalanobis distance. First, obtain the sample sets of the drilling process under normal conditions and the samples to be detected, standardize them respectively to obtain the sample set X and the test set Y, and use X as the reference signal. Secondly, perform spaced augmentation on the samples in X and Y respectively, and calculate the mean vector and covariance matrix of X under different spacings and augmentation times. Thirdly, calculate the Mahalanobis distances between the training samples and the samples to be detected after each spaced augmentation. Finally, perform kernel density estimation on the Mahalanobis distance distribution of the training samples under different spacings and augmentation parameters, set the alarm limit, and compare the Mahalanobis distance of the samples to be detected with the alarm limit to determine whether an abnormality occurs. Finally, conduct experimental verification on the abnormal sticking of the drill, and prove that the present invention can effectively reduce the alarm delay and the false and missed alarm rates when an abnormality occurs, and improve the abnormal monitoring level of the complex geological drilling process.

The invention provides a non-zero mean clutter neutron space signal adaptive detection method and system. The method comprises the following steps: firstly, jointly estimating a mean vector and a covariance matrix of clutters by using to-be-detected data and auxiliary data; then, on the basis of the subspace signal model, deriving a detection statistic by utilizing a generalized likelihood ratio criterion; and finally, comparing the detection statistics with a preset threshold to complete target detection. According to the method, the non-zero mean value of the clutter is explicitly estimated and compensated, so that the adaptive detector keeps a false alarm rate characteristic for a clutter covariance matrix and also has a false alarm characteristic for the non-zero mean value of the clutter in a non-zero mean value clutter environment, and the detection probability is remarkably improved; the technical problem that a traditional subspace detector fails in the environment is solved.

This invention belongs to the field of communications and provides a method and apparatus for diagnosing communication faults, comprising: acquiring a backscattered time-domain signal from a communication optical cable, obtaining multiple intrinsic mode functions through variational mode decomposition, selecting the signal with the largest kurtosis as the analysis signal, reconstructing the phase space of the analysis signal, and determining a set of relevant integral radii; calculating the relevant integral for a preset embedding dimension and candidate delay time values ​​to obtain difference statistics and correlation statistics, constructing a global index function and determining the optimal delay time, extracting the optimal delay time, corresponding curvature, and standard deviation of the difference statistics to form a real-time state feature vector, calculating the Mahalanobis distance based on the normal state benchmark mean vector and covariance matrix, comparing the distance with a preset threshold, and completing the diagnosis of communication optical cable faults.

The invention discloses a prediction method, system and equipment named as a power load prediction method based on a time sequence, relates to the technical field of power prediction, and aims to solve the problems of insufficient date feature expression, non-uniform sample distribution and insufficient description of a model on date sequential logic and semantic association in an existing method. The method comprises the steps of obtaining date type labels of multi-level date types according to historical power loads of dates; obtaining a mean vector and a covariance matrix of each date type as load mode characteristics according to historical loads and date type labels; according to the load mode characteristics, a time sequence prediction model is obtained based on a gating circulation unit, so that rich multi-level date type characteristics and load mode characteristics can be combined; and finally, updating the time sequence prediction model in a rolling manner after the actual power load of one day is newly added every time. Through the scheme, the adaptability and accuracy of power load prediction in a complex and variable load scene are further enhanced.

This invention provides a method and apparatus for blast furnace process fault diagnosis based on dynamic canonical correlation analysis, belonging to the field of blast furnace process fault diagnosis technology. The method includes: acquiring offline and online blast furnace data; performing data identification processing based on the offline and online blast furnace data using a data-driven method to obtain offline zero-mean vectors and online zero-mean vectors; performing fault verification using the SD-CCA method based on a preset fault threshold and the offline and online zero-mean vectors to obtain SD-CCA-based residual vectors and fault detection results; when the fault detection result indicates a fault, performing fault retrieval based on the SD-CCA-based residual vectors and obtaining the fault type based on a fault database. This invention is an accurate and efficient method for blast furnace process fault diagnosis based on dynamic canonical correlation analysis.

The invention provides an anomaly detection method and device, electronic equipment and a storage medium. The method comprises the steps of collecting a data packet obtained by performing database query based on a service code; processing the data packet to obtain index data; determining a corresponding multi-dimensional mean vector and a covariance matrix from a corresponding preset baseline model according to the query statement ID; calculating a comprehensive abnormal score based on the multi-dimensional mean vector, the covariance matrix and the index data; calculating an independent abnormal score based on the index data of each dimension and the historical index data; and if the comprehensive exception score is greater than a preset threshold, generating alarm data based on the comprehensive exception score and the independent exception score of the performance index of each dimension. The method comprises the following steps: establishing index data of a multi-dimensional performance index for each SQL query statement ID; and overall anomaly detection is carried out based on the comprehensive anomaly score determined by the performance index data, so that potential memory overflow and system avalanche risks can be avoided.

The invention provides a bridge influence line identification method and electronic equipment, and relates to the technical field of bridge electric digital data processing. The method comprises the following steps: acquiring bridge response data of a target bridge caused by vehicle movement, and obtaining a bridge quasi-static response based on the bridge response data; acquiring information of vehicles running on a target bridge to construct a vehicle information matrix, and constructing a primary function library; establishing a bridge influence line identification model based on the bridge quasi-static response, the vehicle information matrix and the primary function library; solving a regression coefficient of the bridge influence line recognition model based on a Bayesian algorithm to obtain a posterior covariance matrix and a posterior mean vector of the regression coefficient; and obtaining a bridge influence line identification result based on the posterior covariance matrix and the posterior mean vector. According to the invention, the bridge influence line can be accurately identified.