182 results about "A posteriori probability" patented technology

The embodiment of the invention provides a dynamic risk assessment method and system based on perception and causal reasoning, and belongs to the technical field of automatic monitoring. The method comprises the following steps: acquiring a video data stream in real time and analyzing the video data stream into a structured dynamic visual scene representation; through a neural network model used for executing cross-modal semantic alignment, mapping the dynamic visual scene representation into activation evidence of nodes in a pre-constructed causal knowledge graph in a multi-modal embedding space, the nodes of the graph being risk events or risk factors; and performing probabilistic reasoning on the map based on the activation evidence to calculate the posterior probability of the node corresponding to the to-be-predicted risk event. According to the method, the problems that visual perception and causal reasoning cannot be effectively integrated and the evaluation result lacks interpretability in the prior art are solved, obstacles between visual information and abstract causal concepts are overcome by constructing a perception-mapping-reasoning technology closed loop, and the accuracy, transparency and credibility of risk evaluation are remarkably improved.

The invention discloses a methane column concentration inversion method and system based on a physically-driven variational auto-encoder, relates to the technical field of greenhouse gas monitoring, and is used for solving the technical problem of low inversion reliability of an existing data-driven methane column concentration inversion mode. The methane column concentration inversion method based on the physical driving variational auto-encoder comprises the following steps: pre-training a decoder module through a loss function containing Jacobian matrix consistency constraint, and freezing network weights of all layers of the decoder module; taking the observation spectrum as the input of the encoder module, outputting posterior probability distribution by the encoder module, sampling any concentration value, and inputting the concentration value into the decoder module for forward simulation to generate a simulation spectrum corresponding to the concentration value; when the difference between the simulated spectrum and the observed spectrum is minimum, determining the optimal methane column concentration as an inversion encoder, and inputting the observed spectrum into an encoder module; and outputting the mean value and the standard deviation of the posterior probability distribution.

The invention discloses a bearing fault diagnosis method based on time-frequency double-flow complementation and adaptive gating fusion, and the method comprises the steps: collecting an original vibration signal of a bearing, segmenting an original vibration signal sequence into a plurality of sample segments with fixed lengths, and forming an original data set; dividing the original data set into a training set and a test set by adopting a physical segmentation strategy based on a time axis; constructing a time-frequency double-flow feature extraction module, and extracting a time-domain feature vector and a frequency-domain feature vector; constructing an adaptive gating fusion module to dynamically generate a gating weight, and performing weighted complementary fusion on the time domain feature vector and the frequency domain feature vector to obtain a fused fault feature vector; constructing a classification decision network to map the fusion fault feature vector to a decision space, and optimizing model parameters by using a label smooth regularization loss function and a cosine annealing strategy; and inputting the test set into the trained fault diagnosis model, calculating posterior probability distribution of a sample belonging to each fault category, and outputting a bearing fault diagnosis result.

The invention discloses a household energy centralized management method and device, electronic equipment and a storage medium. The method comprises the following steps: collecting real-time observation data of a family environment to construct a multi-dimensional context feature vector; generating a dynamic state transition probability matrix and emission parameters of the hidden Markov model through a dynamic parameter mapping mechanism, and calculating initial posterior probability distribution of a family activity state by using a forward algorithm; when the information entropy of the probability distribution exceeds a threshold value, executing an active state exploration strategy, issuing an optimal perturbation instruction and correcting a posterior probability based on an active response observation evidence so as to determine a final family activity state; and based on the state loading control constraint set, solving a real-time optimization objective function containing energy efficiency and comfort penalty, and generating an equipment control instruction. According to the method, the uncertainty of state judgment is reduced by combining a dynamic model and an active exploration mechanism, and household energy cost optimization is realized on the premise of ensuring comfort.

The embodiment of the invention discloses a non-contact multi-parameter elbow joint fault diagnosis method and device. Collecting multi-parameter original signals of the area where the elbow-shaped joint is located in the same time reference and the same physical microenvironment; performing denoising, zero offset correction and filtering on the multi-parameter original signal to obtain a pre-processed multi-parameter signal; performing compensation processing on the preprocessed multi-parameter signal by applying a space-time coupling compensation matrix; performing feature extraction on the compensated multi-parameter signals to obtain feature vectors, and calculating response time lag between channels of the parameter signals; calculating a weighted time distance between the typical mode of the current observation stage and each fault stage, and mapping the weighted time distance to a preset fault evolution chain, so as to calculate a posterior probability of the typical mode of each fault stage through the fault evolution chain; and according to a comparison result of the posterior probability and a preset fault threshold value, outputting a fault diagnosis result of the area where the elbow-shaped joint is located and a confidence coefficient corresponding to the fault diagnosis result.

The invention provides a speed limit information fusion method and device, and the method comprises the steps: obtaining map speed limit information and map speed limit confidence based on a map navigation system; acquiring image data in front of a road, and recognizing speed limit information in the image data through a speed limit recognition model to obtain camera speed limit information and camera speed limit confidence; taking the map speed limit confidence coefficient and the camera speed limit confidence coefficient as prior probabilities, and obtaining a posterior probability through a Bayesian algorithm based on the prior probabilities and the conditional probabilities; and the final speed limit information is determined based on the posterior probability, the map speed limit information and the camera speed limit information, so that the accuracy and reliability of the speed limit information are improved.

The invention discloses a single-phase earth fault line selection method based on double-layer parameter adaptive optimization, and the method comprises the following steps: collecting three-phase current corresponding to each feeder line of a target system after the single-phase earth fault of the target system occurs, and carrying out the calculation to obtain zero-sequence transient current; inputting the zero-sequence transient current into a self-adaptive optimization decomposition model for self-adaptive decomposition to obtain an intrinsic mode component of a corresponding feeder line; extracting transient characteristics based on the intrinsic mode component and constructing a transient characteristic matrix; and inputting the transient characteristic matrix into a pre-trained optimization mechanism fault line selection model, outputting a fault posterior probability corresponding to each feeder line, and determining a fault feeder line according to the fault probability. According to the invention, double-layer parameter adaptive optimization is used for carrying out joint optimization on key parameters of a decomposition model and key parameters of a line selection model, so that effective feature extraction and reliable discrimination can still be realized under the conditions of complex zero-sequence transient current high-frequency components and unclear features caused by the access of the distributed power supply; therefore, misjudgment and missed judgment are reduced, and the fault line selection accuracy and robustness are improved.

The invention relates to the technical field of stereoscopic vision, and discloses a synchronous speed visual matching method and system, electronic equipment and a storage medium, and the method comprises the steps: synchronously collecting a stereoscopic video sequence with a predefined frame rate; executing multi-target hybrid tracking and motion induction detection, and outputting target motion information including position and velocity vectors; extracting hierarchical motion features of the target from continuous multiple frames of the stereoscopic video sequence, and performing unified space-time coding; under geometric constraints of stereoscopic vision, scale cosine similarity, direction similarity and trajectory consistency measurement are calculated and serve as observation evidences to be input into the probabilistic reasoning model for fusion, and a posterior probability representing matching reliability is output; the weight distribution of the speed similarity and the direction similarity is adjusted according to the motion characteristics of the targets in the scene, and the stable corresponding matching relation between the left view target and the right view target is established. According to the method, high-time-resolution information can be utilized, motion features and geometric constraints can be effectively fused, and the method has self-adaptive capacity.

The invention discloses a wind turbine generator component life probability prediction method and system, and a wind power plant energy control method and system. The prediction method comprises the steps of constructing a state space model of a degradation process of a wind turbine generator component; acquiring real-time state information of the component; inputting the real-time state information into the state space model, evaluating the degradation state of the component by adopting a sequential data assimilation algorithm, and outputting posterior probability distribution of the degradation state; and based on the posterior probability distribution, forward simulating a future degradation path of the component by adopting Monte Carlo simulation until the future degradation path exceeds a preset failure threshold, and obtaining residual life probability distribution of the component. According to the method, the personalized residual life of each unit can be accurately evaluated, and an innovative wind power plant energy management control strategy is created on the basis, so that the absolute maximization of the full-life-cycle economic benefit of the wind power plant in a complex market environment is realized.

The invention relates to a structure vibration grading early warning method and device and a storage medium, and the method comprises the following steps: obtaining and preprocessing a structure vibration time sequence signal, firstly preprocessing the structure vibration time sequence signal, and dividing a data set; constructing a CNN-LSTM time sequence prediction model to carry out training optimization and short-term prediction; through historical data feature extraction and manifold dimensionality reduction, a vibration state mapping area is determined based on Gaussian mixture model clustering, and a posterior probability is calculated; a classification boundary is determined through maximum likelihood function iteration, and a multi-level early warning threshold value is established; and finally, three-level intelligent early warning of safety, warning and danger is realized by combining the dimension reduction point probability and the threshold value of the prediction data. Compared with the prior art, the intelligent multi-level early warning system is used for realizing high-dimensional time sequence prediction and dimension-reduction multi-level classification early warning in a high-efficiency and high-precision manner, and aims to establish the intelligent multi-level early warning system which can perform structure safety assessment more accurately and more timely.

The invention discloses an efficient identification method for connection rigidity of a steel pipe pole of a power transmission line, which relates to the field of structural health monitoring, and comprises the following steps: constructing a multi-scale finite element model of the steel pipe pole, acquiring displacement and frequency response of the structure to carry out parameter sensitivity analysis, and screening key correction parameters; constructing a key parameter-structure response sample data set, and training and optimizing an FEA-Net neural network model; and constructing a Bayesian correction framework, solving posterior probability distribution by adopting a TMCMC algorithm, and realizing the probabilistic identification of the connection stiffness and the systematic correction of the finite element model. By constructing a local refined contact model, a bolt pretightening force-connection rigidity corresponding relation is established, and an effective means is provided for verifying whether the rigidity of the corrected model is matched with the rigidity of an actual structure or not. The structure rigidity damage position can be accurately recognized, the rigidity degradation degree can be quantified, and a high-credibility model basis is provided for safety evaluation, digital twinborn construction and operation and maintenance decision making of engineering structures such as electric transmission line steel pipe poles and the like.

The invention discloses a data processing method, device and equipment for a QC-LDPC decoder. The method comprises the following steps: storing posterior probability information into a first memory; in the reading stage of the QC-LDPC decoder, soft information corresponding to a layer where a check matrix is located is read from a first memory or a second memory, a check node value corresponding to the layer where the check matrix is located is read from a third memory, and a variable node update value is stored in the second memory; and in a write-in stage of the QC-LDPC decoder, reading the variable node update value from the second memory, storing the check node update value to a third memory, storing the soft information update value to the first memory and the second memory, and storing the soft information update value to the second memory. And the decoding result is read from the second memory after iteration is completed. By optimizing caching and calling of information such as check nodes, variable nodes and soft information, use of memory resources is reduced.

The invention discloses an image definition evaluation method. The method comprises the following steps: acquiring an original image and an optical modulation transfer function; based on the original image and the optical modulation transfer function, generating a multi-scale baseline definition response set, an information-physical definition kernel, an uncertainty measure and an evidence vector; estimating a failure posterior probability based on the evidence vector; generating a gating weight based on the failure posterior probability and the uncertainty measure; and applying the failure posterior probability and the gating weight to execute failure posterior coupling processing on the multi-scale baseline definition response set and the information-physical definition kernel to generate a final definition evaluation result. The coupling process preferably includes applying analytic bias correction and performing cross-scale band selection. According to the method, the evaluation robustness in a complex scene is improved through a failure posterior coupling mechanism.

The invention discloses a control and communication combined maximum posterior probability detection analysis method and device. The method comprises the following steps: calculating and estimating probability density of transmission data according to a control model and a filtering algorithm; performing maximum posterior probability detection according to the probability density and the coding scheme; and analyzing the upper and lower bounds of the maximum posterior probability detection of the control and communication combination according to the probability density and the coding scheme. According to the control and communication combined maximum posterior probability detection and analysis method and device disclosed by the invention, combined detection of communication and control is realized, the transmission error probability can be reduced by utilizing the prior information of the control model, and the transmission reliability and the control performance are improved. According to the invention, the requirements of low-delay and high-reliability control scenes can be effectively met.

This invention discloses an underwater acoustic anti-interference transmission method and system based on sparse time-frequency feature mapping. The method includes: at the transmitting end, adaptively generating a fractional-order linear frequency-modulated waveform with a specific time-frequency shear slope based on the Doppler state of the underwater acoustic channel, achieving physical-layer focusing of transmitted energy; at the receiving end, constructing a hybrid observation model containing wide-block sparse channel components and narrow-block sparse burst noise components, and using a dual-channel variational Bayesian algorithm to jointly iteratively infer the posterior probability distribution of environmental burst noise in the fractional-order domain; finally, recovering the original signal through soft-threshold interference cancellation and fractional-order channel equalization. This invention effectively solves the communication failure problem caused by high-dynamic Doppler diffusion and marine biological impulse noise interference in underwater acoustics by actively mapping the waveform and using heterogeneous sparse joint inference at the receiving end, significantly improving transmission reliability in harsh underwater acoustic environments.

Methods, systems, and devices for enhancing decoding using in-phase and quadrature information. In one aspect, a method includes obtaining in-phase and quadrature values for multiple measurement operations in a quantum error correction code for quantum computing; classifying the measurement results of the plurality of measurement operations using the corresponding in-phase and quadrature values; generating a detector graph of nodes and edges, wherein the detector graph marks detection events occurring in the classified measurements; assigning weights to the edges of the detector graph using a posterior probability distribution of the classified measurements to generate a weighted detector graph; and performing a decoding process on the weighted detector graph to calculate a decoded output of the decoding process, where the decoded output predicts an occurrence of an error in the quantum computation.

Disclosed is a natural language interaction-based method for extracting a parameter of an energy storage management system, belonging to the technical field of energy storage of new energy power systems. The method comprises parameter type division, natural language interaction corpus collection, parameter labeling, construction of a word segmentation set, natural language interaction corpus encoding, computation of a prior probability and a conditional probability, persistence of the prior probability and the conditional probability, new corpus input, loading of the prior probability and the conditional probability, computation of a posterior probability on the basis of a naive Bayes formula, and extraction of a parameter corresponding to the maximum value of the posterior probability.

The application discloses a site micro-vibration vibration source identification method based on machine learning, signals measured by acceleration sensors at the edge of a measured site and acceleration sensors at the center of the measured site are processed in sequence, including denoising, transient impact signal extraction, feature extraction, clustering processing of dynamic time warping distance of the transient impact signal based on a feature matrix, GMM modeling, obtaining a model, verifying the number of independent Gaussian components of the model, selecting the number of components with the minimum value as the optimal model according to the Akaike information criterion value, and performing maximum a posteriori probability identification on the transient impact signals captured by the sensors; GMM parameter comparison is performed on sets that may belong to the same category, two sets with highly overlapped models are merged into the same set, and the set is used as vibration data sets generated around the site and having an impact on the center position of the site.

The application discloses a kind of based on crowdsourcing data to the method, device and equipment of extracting ground feature element, and relates to map data technical field.The method comprises the following steps: obtaining first identification result;First identification result is the result obtained by identifying the information collected by first collection device at target position at first time;When first identification result indicates that ground feature element is identified in target position based on the collection information of first collection device, obtain first preset condition probability parameter and first prior probability of first collection device;According to first preset condition probability parameter and first prior probability, first posterior probability is calculated;When first posterior probability is not less than first threshold, ground feature element is extracted.The application can timely and accurately extract ground feature element, so as to update map timely and accurately.

The invention provides a construction deformation prediction method and device, electronic equipment and a storage medium, and belongs to the technical field of data prediction, and the method comprises the steps: obtaining historical working condition parameters of a subway tunnel construction process; constructing a regularization item related to a sample point feature space; determining a decision tree splitting criterion based on a regularization item, recursively dividing each feature in the sample point feature space into non-overlapping hyperrectangular regions, and constructing a decision tree; calculating a posterior probability distribution parameter of each leaf node in the decision tree based on a Bayesian inference updating rule; defining a structure risk penalty function in combination with the posterior probability distribution parameters to perform post pruning on the decision tree to obtain an optimal decision tree; real-time geological exploration parameters and real-time shield tunneling machine tunneling parameters in the subway tunnel construction process are collected; and real-time geological exploration parameters and real-time shield tunneling machine tunneling parameters are input into the optimal decision tree, and a monitoring point ground settlement prediction interval is output, so that the deformation prediction accuracy under the complex geological condition is effectively improved.

This invention belongs to the field of visible light communication technology and relates to a nonlinear blind equalization method for a VLC system based on a variational autoencoder. The method includes: inputting the received signal into a trained blind equalization network model to obtain an equalized signal; the training process of the blind equalization network model includes: preprocessing the transmitted and received signals of the VLC to obtain a preprocessed transmitted signal Tx and its counterpart to the received signal Rx; and inputting Rx into an equalizer network to obtain the latent variable signal Tx. * The prior probability distribution and the posterior probability distribution of a given received signal Rx are obtained; the prior probability distribution and the posterior probability distribution are input into the channel estimation network to obtain the likelihood probability distribution and the equalization signal; the loss function value is calculated based on the prior probability distribution, the posterior probability distribution and the likelihood probability distribution, and the parameters of the model are updated based on the loss function value until the stopping condition is reached; the present invention integrates multiple loss functions to avoid the local optimum problem caused by a single loss function.

A cloud environment fault handling method, executed by an electronic device, and comprising: acquiring current fault alarm information of a current cloud environment, wherein the current fault alarm information comprises at least one piece of operating state data (101); acquiring a pre-stored fault alarm information set, wherein the fault alarm information set comprises at least one item of historical fault alarm information of the current cloud environment (102); acquiring a conditional probability set and a prior probability set corresponding to the historical fault alarm information, wherein the conditional probability set comprises conditional probabilities corresponding to the categories of respective pieces of operating state data, and the prior probability set comprises prior probabilities corresponding to respective preset fault categories (103); on the basis of the category of the operating state data, screening the conditional probability set for a conditional probability corresponding to the operating state data (104); on the basis of the conditional probability and the prior probabilities, determining posterior probabilities that the current fault alarm information belongs to the respective preset fault categories (105); and on the basis of the posterior probabilities, screening the preset fault categories for a selected fault category corresponding to the current fault alarm information, and sending the current fault alarm information to a service server corresponding to the selected fault category (106).

The application discloses a semantic recognition word sequence determination method and device, a storage medium and an electronic device, and relates to the technical field of smart homes. The method comprises the following steps: performing linear projection on an implicit vector sequence corresponding to a voice signal to obtain a multi-granularity posterior probability of the voice signal, wherein the multi-granularity posterior probability comprises a character posterior probability, a subword posterior probability and a word posterior probability; searching for a word set and a character set associated with a subword in the voice signal in a multi-level mapping table, and calculating a target path score of the subword in the voice signal according to the multi-granularity posterior probability, the word set and the character set; inputting the target path score into a target decoder to obtain a target subword sequence, and obtaining a target word sequence corresponding to the target subword sequence according to the multi-level mapping table. Through the application, the problem of homonym misrecognition of rare words, multi-sound words and professional terms in the process of end-to-end ASR in related technologies can be solved.

The invention is suitable for the technical field of crack monitoring, and relates to a crack identification method based on acoustic emission, a medium and a terminal.The method comprises the steps that S10, RA parameters and AF parameters of AE signals are extracted, and a two-dimensional feature set is constructed; s20, training a GMM model by adopting an EM algorithm, and outputting a mean value, a covariance matrix and a weight of each Gaussian distribution component; s30, determining the posterior probability of each AE event in all Gaussian distributions; s40, comparing the posterior probability of each component, and taking the category corresponding to the maximum value as the classification result of the crack type of the event; s50, introducing a crack identification confidence index Ci, and selecting a threshold Cth to judge a result; and S60, introducing a crack mechanism energy dominance index for judging the energy dominance degree of a shear crack mechanism in the failure process, and assisting engineering analysis. The method is simple in process and convenient to operate, and the accuracy of crack identification is effectively improved.

The application discloses a robust multi-sequence reconstruction method based on maximum posterior probability in DNA storage, and comprises the following steps: decoding each sequence in a cluster by using an improved BCJR decoder; converting each sequence in the cluster into corresponding time sequence representation; determining the weight of each sequence; and deducing a MAP algorithm formula of robust multi-sequence decoding, and integrating the sequence weight into joint posterior probability calculation to inhibit the influence of outlying sequences. By proposing a base IDS channel model, the BCJR decoding algorithm is improved, so that the defects that the existing statistical inference sequence reconstruction method is only suitable for binary error correction code are overcome. Therefore, the application can process the quaternary error correction of the base form in the DNA sequence.

This invention discloses an adaptive tracking method and device for anti-dropped decoys based on a multi-model approach to measurement errors, belonging to the field of radar target tracking technology. The method includes constructing motion and centroid interference models for passive radar dropped decoys to determine the impact of interference on radar measurements; constructing a set of passive radar measurement models characterizing the time-varying characteristics of measurement noise, and establishing a hybrid system with uncertain measurement noise by combining the target motion model; employing an interactive multi-model architecture combined with a capacitive Kalman filter algorithm to perform interactive and parallel filtering on each sub-model of the hybrid system; updating the posterior probability of the measurement noise in each sub-model at the current moment in real time based on the measurement residuals and information covariance; and performing weighted fusion of the state estimates of each sub-model to obtain the adaptive tracking result. This invention effectively solves the tracking divergence problem caused by weak observability and time-varying measurement errors in passive radar under dropped decoy interference, significantly improving the tracking accuracy and robustness of the system.

There are provided a method and a system for extracting singing voice phoneme duration. A singing voice phoneme duration extraction system using a MIDI according to an embodiment may receive phonemes converted from a text as input, and may output a prior probability distribution, may receive acoustic features as input and may output a posterior probability distribution, may convert the probability distribution, may perform monotonic alignment search by using information on MIDI duration, and may output a waveform which is a voice digital signal, based on input reflecting a result of extracting the phoneme duration.

The embodiment of the invention discloses a network quality anomaly detection method and system. The method comprises the following steps: acquiring time sequence data of a network quality index, and preprocessing the time sequence data by using an encoder to obtain hidden state vectors at a plurality of moments; inputting an initial global context vector obtained according to the hidden state vector into a preset hidden Markov model to obtain a hidden state posterior probability vector at each moment; inputting the hidden state vector and the hidden state posterior probability vector into a preset attention layer, outputting an attention weight of each moment by the attention layer, and generating a target global context vector according to the attention weight and the hidden state vector; and inputting the target global context vector into the hidden Markov model to obtain a network quality anomaly detection result. According to the technical scheme of the embodiment of the invention, the network quality anomaly detection precision can be improved.

The invention provides an optimization method and device for a sensor array, and the method comprises the steps: building a linear array model, carrying out the discretization of a parameter space of an array element position, and generating an array arrangement position grid; acquiring a plurality of sampling points in a sampling space of a preset visual area; constructing a first sensing matrix according to the array position grid and the plurality of sampling points, and establishing a sampling signal model containing sparse signals according to the first sensing matrix; based on a preset multilayer prior model and a posterior probability density function, determining a posterior probability mean value of the sparse signals, and updating the sparse signals in the sampling signal model according to the posterior probability mean value; and performing linearization processing on the compressed sensing vector deviation corresponding to the updated sampling signal model through first-order Taylor approximation, constructing a target gradient matrix, correcting the first sensing matrix according to the target gradient matrix to obtain a second sensing matrix, and obtaining a target array arrangement position according to the second sensing matrix.

The application specifically relates to a passive synthetic aperture sonar direction finding method based on variational message passing, and belongs to the technical field of signal processing. The application adopts a broken line model to fit a distorted array shape, and establishes an array receiving data model containing array shape disturbance, in view of the problems that a traditional algorithm can only process space-time coherent signals, is sensitive to array shape distortion, and has high operation complexity; a spatial domain super-complete dictionary is constructed to convert a direction finding problem into a sparse reconstruction problem, a layered probability model is established by using a Gaussian scale mixed model and is expressed as a factor graph; each parameter posterior probability is solved by using variational message passing iteration, high-dimensional matrix inversion is avoided, and angle fine estimation is realized by combining Laplace interpolation. The application can jointly estimate a signal source direction and array shape distortion parameters, keeps high precision in a random signal and distorted array shape environment, significantly reduces operation complexity, and is suitable for underwater passive synthetic aperture high-resolution direction finding engineering application.