17 results about "Joint likelihood" patented technology

This application discloses a method for association analysis based on high-throughput dynamic phenotypes and genetic effects, relating to the field of agricultural bioinformatics. By fitting growth curves to high-throughput dynamic phenotype data from multiple time points, the dynamic changes in the growth cycle of a biological population are obtained. A joint likelihood function algorithm is then constructed to associate this dynamic vector with genetic effects, revealing genetic markers related to growth and development trends and processes. Compared to existing GWAS methods, which can only analyze static, single-time-point phenotype data and cannot uncover genetic loci regulating biological growth and development, this application, by performing association analysis on data collected from multiple consecutive time points, can more comprehensively reveal the evolutionary patterns of phenotypes over time, thereby achieving accurate analysis of genetic markers.

The invention belongs to the technical field of intelligent manufacturing and industrial Internet of Things, and provides an automobile welding workshop early warning method and system based on historical and real-time data, and the method comprises the steps: constructing phase sensing features of different scales based on a collected phase control signal of a welding process; based on the phase perception characteristics of different scales and the constructed phase perception depth time sequence prediction model, a process parameter prediction value is obtained; extracting causal relationship prior information based on a phase perception depth time sequence prediction model, mapping the extracted causal relationship prior information into an edge prior probability of a Bayesian network structure, and constructing a hierarchical Bayesian network of prediction-diagnosis bidirectional coupling according to the edge prior probability; and calculating a joint likelihood according to the real-time observation value and a predicted value of the phase perception depth time sequence prediction model, obtaining a joint likelihood anomaly probability in combination with an anomaly posterior probability deduced by a hierarchical Bayesian network, and generating an anomaly trigger decision strategy in combination with the joint likelihood anomaly probability and a joint likelihood anomaly criterion. And the early warning accuracy is improved.

The present application relates to radar target tracking technical field, specifically to a kind of for radar weak target's detection front tracking method.For simultaneously estimating target state and measurement noise covariance, first need to calculate the joint probability density of target state and measurement noise covariance.Then introduce Gaussian inverse gamma mixture distribution to model joint probability density, since target state and measurement noise covariance are coupled in joint likelihood function, this will lead to joint posterior probability density difficult to solve analytically, therefore, approximate solution of separable approximation of joint posterior probability density is solved using variational bayesian method.Finally, in filtering update phase, based on the separable approximation solution, information exchange is performed, i.e., each Bernoulli component uses the predicted state information shared by other Bernoulli components to perform update.Simulation verification shows that, in low signal-to-noise ratio scenario, the present application can adaptively estimate measurement noise covariance, and tracking accuracy is improved.

The application provides a kind of unmanned vehicle wheel ground coupling model optimization method and system based on machine learning, it is related to simulation technical field, the method comprises: by dividing model variable parameter and calibration parameter, construct simulation model, obtain simulation data set using Latin hypercube sampling, combined with preset driving condition to carry out real vehicle test and collect experimental data set;Gaussian process proxy model of simulation model is constructed to replace simulation model, Gaussian process model of modeling deviation function is built for modeling deviation, and data correlation is established;Integrate simulation data set and experimental data set, build joint Gaussian process model, calibration parameter calibration and model deviation correction are simultaneously completed by maximizing joint likelihood function, finally output response prediction value, the application does not need to rely on parameter prior distribution, distinguishes quantization modeling deviation and parameter deviation, greatly improves model precision and generalization ability, optimizes calculation efficiency.

The invention discloses a group target joint tracking and classification method based on active and passive radar interactive fusion, which belongs to the field of multi-sensor fusion, and comprises the following steps of: modeling active and passive radar measurement by using a double-layer random set, so that the active and passive radar measurement can be filtered under a unified Bayesian theory framework; passive measurement is processed through a fuzzy measurement filtering method, and the position and category information of the passive radar about the target is obtained; establishing a joint likelihood function of active and passive radar heterogeneous measurement by using a random set so as to construct a coupling relationship between target tracking and classification; on the basis of a centralized fusion method, active and passive radars are organically combined, filtering of heterogeneous measurement information is completed, position and category joint posterior distribution is obtained, and target tracking and classification integration is achieved. The method has the characteristics of low calculation complexity, high tracking and classification precision, strong algorithm robustness and the like.

The invention discloses an atmospheric pollution source tracking method and system based on big data, and relates to the technical field of atmospheric pollution prevention and control, and the method comprises the steps: collecting multi-source data, carrying out the optimization of the collected data through a generalized additive model, and generating a fusion data set; obtaining a corresponding feature map through a CNN model and a positive definite matrix factorization method based on the fusion data set, generating a fusion tensor, obtaining a concentration contribution coefficient through a Gaussian plume model, constructing a joint likelihood function, constructing Bayesian prior based on the fusion tensor and the joint likelihood function, and generating a visual map layer; and identifying and sorting emission hotspots based on the visual layer to obtain an emission hotspot list, obtaining a key area pollution report through the emission hotspot list, generating a pollution source intervention strategy report based on the key area pollution report, and performing strategy execution effect evaluation. According to the invention, the source positioning precision of atmospheric pollution source tracking is improved.

The invention provides an unmanned wheel ground coupling model optimization method and system based on machine learning, and relates to the technical field of analog simulation, and the method comprises the steps: building a simulation model through dividing model variable parameters and calibration parameters, employing Latin hypercube sampling to obtain a simulation data set, and obtaining a simulation data set; carrying out a real vehicle test in combination with a preset driving condition to collect an experimental data set; constructing a Gaussian process agent model of the simulation model to replace the simulation model, constructing a Gaussian process model of a modeling deviation function for the modeling deviation, and establishing data association; a simulation data set and an experiment data set are integrated, a joint Gaussian process model is built, calibration parameter calibration and model deviation correction are synchronously completed by maximizing a joint likelihood function, and finally a response prediction value is output. The method does not need to depend on parameter prior distribution, quantized modeling deviation and parameter deviation are distinguished, the model precision and generalization ability are greatly improved, and the method is suitable for large-scale popularization and application. And the calculation efficiency is optimized.

The invention discloses a data-driven demand response feature recognition and uncertainty quantification method in a power system, and the method comprises the steps: firstly collecting the historical demand response data of a user, and constructing a data set comprising a user feature vector X, a response intention label Y and a response potential label Z; and establishing a mapping relation between the user feature vector X and the response will probability by using a generalized linear model, and assuming that Y obeys Bernoulli distribution. Aiming at a user with a response intention, establishing a mapping relation between conditional expectation of a response potential label Z and X through a generalized linear model framework, and constructing a parameterized probability distribution model of the Z; on the basis, joint probability distribution of Y and Z is constructed. And based on a sample independent identically distributed hypothesis, constructing a logarithm joint likelihood function, obtaining optimal parameter estimation by adopting a gradient descent algorithm, and determining an optimal model through AIC and BIC criteria. And finally, quantifying the uncertainty of response willingness and potential, and constructing a comprehensive index to evaluate the reliability degree of user response.

This application relates to a multi-source sensor fusion localization method and system for partial discharge in primary equipment of a substation. The method includes: constructing a digital twin model based on the equipment's three-dimensional geometric parameters, electromagnetic properties of the internal medium, and acoustic properties; acquiring ultra-high frequency electromagnetic wave signals and acoustic emission signals and performing multipath decomposition to obtain electromagnetic wave multipath delay cluster feature vectors and acoustic wave multipath delay cluster feature vectors; calculating path mismatch degrees for all hypothetical source point locations based on the feature vectors to obtain a path dependency matrix; calculating a wave velocity collaborative correction factor for each hypothetical source point location based on this matrix, and performing joint likelihood estimation by combining electromagnetic wave and acoustic wave path mismatch degrees to obtain a preliminary spatial likelihood distribution; and performing variational Bayesian inference on this distribution to obtain the localization result and confidence interval evaluation index. This method can effectively eliminate false localization points and significantly improve the accuracy and reliability of partial discharge localization in complex structural equipment.

The invention discloses an acceleration coefficient interval estimation system, which comprises the following steps of: acquiring sufficient complete failure sample data through a Testto-Failure test, designing a'stress layer and degradation characteristic layer 'double-layered Bootstrap sampling rule, and generating a plurality of groups of failure sample combinations; meanwhile, whole-course degradation data of failure samples under double stress are integrated, a cross-stress integrated joint likelihood function is constructed, the acceleration coefficient of each combination is directly solved, and finally the acceleration coefficient interval is determined through statistical analysis. By the adoption of the technical scheme, the problems that existing acceleration coefficient estimation can only achieve point estimation and lacks effective interval estimation adaptive to Testto-Failure data features, an interval result cannot cover failure scene parameter fluctuation, and error accumulation exists in traditional step-by-step solving are solved, and the accuracy of acceleration coefficient estimation and the interval reference value are improved.

The invention discloses a spatial multi-omics analysis method based on a probability framework and a generalized principal component analysis principle, and aims at the core challenges of high dimension, modal heterogeneity and spatial complexity of spatial multi-omics data to realize efficient integration and accurate spatial analysis of multi-modal data. Preprocessing the data to complete feature screening and normalization, and converting the data into a matrix format; in model training, independent space Gaussian process priori is distributed for each potential factor, a weighted joint likelihood function is adopted to realize multi-modal collaborative modeling, efficient parameter inference is completed through step-by-step iterative optimization, and a low-dimensional potential factor matrix integrating space information and multi-modal features is output; standardized biological analysis such as spatial domain detection, clustering quantitative evaluation, spatial visualization, difference feature screening, function enrichment and cell interaction analysis is realized.

The application discloses a petrochemical device protection layer failure number prediction method, comprising the following steps: S110, acquiring key safety variables and protection layers of the device; S120, collecting real-time monitoring values of the key safety variables and protection layer trigger states of the device, and counting failure numbers of each protection layer in each period; S130, calculating joint likelihood distribution functions of all protection layers; S140, calculating posterior failure number distribution functions of each protection layer according to the Bayes theory; and S150, dynamically correcting the failure numbers of the protection layers in the next period by using a Markov chain method. The application further discloses a petrochemical device protection layer failure number prediction system, an electronic device and a storage medium. The application monitors fluctuation conditions of the protection layers of the key safety variables of the device in a period, counts the failure numbers of the protection layers in the period, and dynamically predicts the failure numbers of the protection layers in the next period based on the Bayes algorithm and the Markov chain method, thereby providing technical support for device-side dynamic risk early warning.

The application discloses a position error correction method based on maximum likelihood estimation, comprising the following steps: acquiring radar error positions of each radar, positions of a correction source at each moment and distances between the radar and the correction source at each moment; obtaining a joint likelihood function of the distance between the radar and the correction source and the radar error position based on a joint likelihood function of the radar error positions of each radar and a joint likelihood function of the distances between each radar and the correction source at each moment; making the joint likelihood function of the distance between the radar and the correction source and the radar error position maximum to obtain a target function which needs to be minimized; and processing the target function by using a particle swarm algorithm to obtain a corrected position of the radar. The method of the application corrects the radar position by using a mobile correction source, does not need the accurate position of the correction source, corrects the radar position error by using the distances between each radar and the correction source at different moments and the error positions and the joint likelihood function of the distance between the radar and the correction source and the radar error position.

The invention discloses a Bayesian filtering prediction method for the stress-strain relationship of a geotechnical material, and relates to the technical field of geotechnical engineering, and the method comprises the steps: obtaining the actual measurement macroscopic response of a target geotechnical material under a set working condition, the actual measurement macroscopic response being macroscopic stress-strain data obtained through a test, and microstructure response data; based on macroscopic stress-strain test data and microstructure response data, determining an initial value range of a nano-scale GB potential parameter and a micro-scale discrete element parameter, generating an initial multi-scale model parameter set, and constructing a joint likelihood function; based on a joint likelihood function, introducing a Bayesian filtering framework, and performing posteriori updating on the initial multi-scale model parameter set to obtain parameter posteriori distribution; selecting sample points from the parameter posterior distribution, calling discrete element simulation for each sample point to obtain a corresponding macroscopic stress-strain response, forming a training data set, and constructing a Gaussian process proxy model based on the training data set.

The invention relates to a question and answer quality evaluation model training method and a question and answer quality evaluation method. Constructing a joint likelihood function according to the noise probability function and the preference probability function of the review model corresponding to all the group question and answer pairs; according to the joint likelihood function, constructing a target loss function corresponding to the question and answer quality evaluation model to be trained; training a to-be-trained question and answer quality evaluation model according to the question and answer pair preference data set, and iteratively updating trainable model parameters and to-be-estimated reliability parameters corresponding to each preset evaluation model in the training process until the target loss function converges, so as to obtain the question and answer quality evaluation model; the joint optimization of the trainable model parameters and the to-be-estimated reliability parameters of the preset review model is realized in the absence of absolute labels, and the robustness, generalization ability and training transparency of the model are improved.

The present invention belongs to the field of robotics and discloses a feature fusion and refinement embedded sparse Bayesian learning method and system for robot surface milling contour error monitoring. The feature vectors composed of process parameters, robot stiffness, cutting force and tracking error are structured and fused into a dictionary matrix, and combined with the sparse Bayesian learning method to realize robot processing error monitoring. During model training, the present invention can obtain the posterior distribution of weight variables and estimate the expectation and variance of the posterior distribution of weight coefficients by maximizing the posterior probability. The variance and hyperparameters affecting the sparsity of the weights are estimated by maximizing the marginal likelihood function, that is, maximizing the expectation of the joint likelihood function under the posterior distribution of the weights to optimize. The minimum solution of the converted loss function can be obtained to optimize the hyperparameters. The posterior expectation and variance of the model weights, as well as the hyperparameters, are iteratively updated during the training process, and the iterative process is terminated according to a set change threshold to achieve the sparsification of the model weight vector and the construction of a sparse model. The present invention realizes robot milling processing error monitoring of parts with different characteristic surfaces under variable postures, effectively reducing the part measurement cost.

The present application relates to the field of geotechnical engineering parameters, in particular to a method and system for quickly determining soil strength parameters based on CPT data. The method comprises: obtaining CPT data and drilling data, and obtaining standardized data; constructing a local neighbor sample set according to the spatial relationship between the target CPT query point and the drilling point, and obtaining the KNN weighted representation value; fitting the edge distribution of the KNN weighted representation value of cohesion and internal friction angle respectively, selecting the optimal edge distribution and constructing the prior distribution; constructing the normalized CPT response according to the standardized CPT data, and constructing the main CPT likelihood function; constructing the shear strength weak constraint likelihood function by using the KNN weighted representation value of shear strength; constructing the joint likelihood function based on the two kinds of likelihood functions, and updating the prior distribution to the posterior distribution combined with the Bayesian update, and outputting the posterior parameter and uncertainty evaluation result. The present application solves the problems of insufficient expression of CPT empirical estimation uncertainty and difficulty of updating the target CPT observation with drilling information in the prior art.