8 results about "Bayesian Prediction" patented technology

The invention relates to the field of marine environment data processing and analysis, and discloses a coral whitening risk monitoring method based on multi-source environment information, and the method comprises the steps: obtaining the multi-source environment information of a target monitoring sea area in a historical period; performing correlation analysis on the multi-source environment information in the historical time period, and calculating a correlation coefficient between a weekly heat index and each marine environment parameter; when a correlation coefficient between the weekly heat index and any marine environment parameter is greater than or equal to a preset coefficient threshold value, removing the corresponding marine environment parameter from the multi-source environment information to obtain screened multi-source environment information; constructing a Bayesian prediction model according to the screened multi-source environment information; the Bayesian prediction model comprises fixed effect items, and the fixed effect items comprise independent action items and interaction action items; and based on the Bayesian prediction model, calculating the coral whitening probability of the target monitoring sea area in the to-be-monitored time period. According to the invention, the accuracy and reliability of the coral whitening risk monitoring and early warning result can be improved.

The application provides a dangerous area recognition method and device based on machine vision, and relates to the technical field of dangerous area recognition.The application obtains monitoring image data of a danger source in a historical monitoring area, forms a training image set through preprocessing, constructs a danger recognition model, and trains the danger recognition model, performs danger source recognition on real-time monitoring image data according to the trained danger recognition model, demarcates a basic danger area according to a basic safety distance set according to the recognition result, calculates a danger coefficient of each type of static area according to historical data of each type of static area, adds an early warning distance, calculates a dynamic danger coefficient according to historical data of each type of dynamic equipment, calculates a buffer distance in front of a moving direction, obtains historical danger accident data of a to-be-recognized area, extracts an equipment feature set, and identifies a future safety state of a sub-area through a Bayesian prediction model.

The application provides a new energy mine truck thrust rod fatigue life prediction method, belonging to the field of federated learning and application technology. First, a multi-mine area multi-source fatigue feature dataset is constructed. Then, a double-layer neural network structure based on Bayesian modeling is designed, including a sparse prior generation module, a thrust rod time series feature encoding module and a thrust rod residual life Bayesian prediction module, to realize model parameter sparsification, time series feature expression and uncertainty quantification. Further, a graph structure based federated training and information aggregation mechanism is adopted, combined with global aggregation and local graph modeling strategy, to realize knowledge sharing and personalized adaptation between different mine areas. Finally, an online fine-tuning mechanism based on uncertainty driving is proposed, realizing rapid adaptive optimization of the global model in the new mine area environment through Bayesian inference. The method can significantly improve the accuracy, robustness and cross-domain generalization ability of thrust rod life prediction while ensuring data privacy.

This invention discloses a machine learning-based method for predicting the compressive strength of geopolymer-stabilized soil, belonging to the field of building material prediction technology. The method includes: constructing a Bayesian prediction model for geopolymer properties based on theoretical strength values, combining a physical constraint layer and a Bayesian probabilistic inference layer; training the Bayesian prediction model to generate a trained model; inputting the mix proportion parameter vector of the geopolymer-stabilized soil into the trained model to perform multiple Monte Carlo sampling predictions to obtain predicted strength values ​​and physical constraint strength values; and performing statistical analysis on the predicted strength values ​​to generate prediction confidence intervals. This invention, by generating prediction confidence intervals and physical contribution parameters, achieves a simultaneous characterization of the distribution characteristics of predicted strength and the degree of mechanistic influence, and realizes a unified expression of uncertainty quantification and mechanistic contribution within a machine learning framework.

The invention discloses a wind turbine generator yaw error prediction control method and system based on deep learning, and the method comprises the steps: firstly processing multi-dimensional original sensor data through a self-adaptive denoising module, separating out a pure physical signal, and obtaining an instantaneous difference signal; furthermore, the difference signal is modeled in combination with the operation context of the unit, and the expected noise characteristic under the current working condition is predicted, so that the attributive noise level estimation is generated, and the normal operation fluctuation and the abnormal interference are effectively distinguished. On the basis, de-noising input and noise level estimation are cooperatively input into a Bayesian prediction model, prediction yaw error distribution containing a mean value sequence and a variance sequence is output, and explicit quantization of prediction result uncertainty is achieved. Finally, control risk assessment is performed based on the distribution, and adaptive control parameters are dynamically synthesized to optimize yaw action, so that the accuracy and safety of a control decision in an uncertain environment are ensured while data noise interference is suppressed.

The invention discloses a power distribution switch state monitoring method, a power distribution terminal, a system and a medium, and the method comprises the steps: collecting an electrical quantity waveform, a mechanism DC waveform and the working condition data of a power distribution switch in response to a switching operation event of an operation mechanism on the power distribution switch; inputting physical characteristics extracted and corrected from the electrical quantity waveform and the mechanism direct current waveform into the LSTAN, outputting a single operation health score and a fault mode probability vector of the operation mechanism, and updating a health index based on the single operation health score; using a Bayesian prediction model to predict the residual life of the operating mechanism based on the accumulated operation frequency and the health index of the operating mechanism; and performing safety maintenance on the distribution switch based on the single operation health score, the fault mode probability vector, the health index and the residual life. The multi-dimensional features of the power distribution switch are collected through the power distribution terminal, the state of the power distribution switch is monitored in time based on the local model, and compared with an existing scheme, the state of the power distribution switch can be sensed in time and reliably.

The wind turbine blade residual life prediction method based on the unscented Kalman filtering disclosed by the application comprises the following steps: S1, a blade damage model is established, a wind turbine is operated, and cracks begin to appear on the blade and gradually expand; S2, the blade crack expansion is converted into a discrete cumulative process to obtain a state transition equation and an observation equation; S3, the annual average wind speed, the cut-in wind speed, the cut-out wind speed and the Rayleigh distribution are determined; S4, the observation data of the blade crack length are obtained; S5, the damage state process is defined, and the covariance matrix of the system noise and the crack length measurement error is set; S6, the Bayesian prediction is performed by using the unscented Kalman filtering method, and the Bayesian update is performed; S7, the blade crack expansion damage threshold is set, whether the updated state exceeds the set damage threshold is judged, and the residual service life is calculated. The prediction method solves the problem that it is difficult to coordinate the error reduction and the calculation complexity in the prior art.

The invention discloses a coal cutter navigation cutting planning method and system based on a working face step model, and the method comprises the steps: directly representing the step mutation of a coal seam floor through constructing a discretized working face step model; dividing a three-level geological cognitive state based on the completed cutter number and the predicted sight distance, realizing spatial quantization of geological information uncertainty, and establishing a credible geological model dynamic evolution mechanism from prior driving to data driving; on this basis, the floor height is modeled as a first-order Markov hidden state variable, probability distribution of a future floor profile is generated by adopting Kalman filtering and multi-step Bayesian prediction, and online compensation is carried out by combining exponential decay weighting historical deviation; and finally, by taking the prediction profile as a main track and the height adjustment compensation amount output by the cutting current layering system as feedback correction, generating the target height of the roller through adaptive weighted fusion, and realizing collaborative optimization of geological cognition, multi-step prediction and autonomous height adjustment.