15results about How to "Resolve uncertainty" patented technology

The application discloses an expression recognition method and system based on uncertainty learning, comprising the following steps: S1, obtaining a face image to be recognized; S2, preprocessing the input face image to obtain a standardized face image; S3, initializing a deep learning training condition; S4, training an expression recognition model by using a plurality of modules, including a face feature extraction module, a multi-dimensional feature classification module, a latent label distribution learning module and an attention consistency module, and updating model weight parameters; and S5, obtaining an output expression category of the face image through the expression recognition model. The expression recognition method and system based on uncertainty learning solve the uncertainty problem of face expressions in a wild environment by combining a latent label distribution and an attention consistency method, improve the accuracy of expression recognition by mining the latent label distribution, and enhance the interpretability of the model by using a generated attention map method.

The application discloses a sealing ring position distribution description method under a multi-scale assembly factor of a flared joint, establishes a multi-scale manufacturing and assembly factor fusion model containing a matching angle, a radial circle runout and a half-flared angle; a reference coordinate system is established with the center of a circle intersecting a tapered surface and a cylindrical surface in the flared pipe as an origin, a macroscopic matching angle is represented by rigid body rotation, a mesoscopic radial circle runout is represented by a radius of a point on a generatrix of the tapered surface of the pipe joint, and a mesoscopic half-flared angle is represented based on a spherical coordinate; a coordinate of a contact point at the top of the edge of the flared pipe is located, a local analytical model is established to represent an intermediate parameter angle changing with the matching angle, and finally, a relative contact position of a contact area of the flared pipe and the pipe joint is determined as an initial boundary condition of a microcosmic finite element contact model. The application can provide an accurate initial contact state and boundary condition for a contact model with a microcosmic morphology, and improve the prediction accuracy of sealing performance.

The application provides a kind of shipboard aviation equipment composite material damage D-S evidence theory nondestructive evaluation method, belongs to the field of shipboard aviation equipment nondestructive evaluation, it includes the following steps: at time, the triaxial damage area characteristics and triaxial vibration frequency characteristics of the composite material structure of shipboard aviation equipment are synchronously acquired and stored;The triaxial damage area characteristics and triaxial vibration frequency characteristics at time are compared respectively, and the corresponding triaxial damage area change degree and triaxial vibration frequency change degree are obtained;The identification framework of D-S evidence theory is established, and the triaxial mass function of the composite material structure is calculated respectively;According to the created D-S damage judgment model, the classification judgment and early warning of the composite material damage of shipboard aviation equipment are carried out.The application can realize the real-time classification evaluation of the dynamic damage of the composite material component of shipboard aviation equipment, and provide data support for the condition-based maintenance of shipboard aviation equipment.

The application discloses a self-adaptive multi-modal data fusion bridge health monitoring method and system, generates a response sample database through establishment of a bridge nonlinear dynamics equation, pre-processes bridge sensor data and trains a deep learning model to extract damage features, and self-adaptively optimizes model parameters; based on model output, real-time and historical data are combined to perform bridge health evaluation, and a monitoring strategy is dynamically adjusted or early warning is triggered; an edge computing device is deployed to perform pre-processing and damage identification, and long-term trend analysis is realized in cooperation with a cloud platform; macro-structure damage and micro-crack data are fused through cross-scale feature extraction; an artificial intelligence algorithm is used to generate a bridge health evaluation result, a bridge maintenance and repair scheme and a decision-level early warning signal. The application combines nonlinear dynamics modeling, self-adaptive deep learning, edge cloud collaborative computing and cross-scale analysis, improves complex damage identification precision and monitoring intelligent level, and is suitable for large-scale traffic network bridge monitoring management.

The invention provides a polarization image defogging method based on a neural network, and belongs to the technical field of image processing, and the method comprises the steps: obtaining a plurality of fogged polarization images, and generating a polarization image package; constructing a neural network defogging model; and based on the constructed neural network defogging model, carrying out defogging processing on the foggy polarization image in the generated polarization image packet to obtain a fogless image. According to the image defogging method, the deep learning and the polarization imaging technology are combined, the problem that the calculation complexity of the polarization imaging technology is high can be effectively solved, the overall generalization ability and defogging level are improved, and better robustness and high adaptability can be shown in the face of different types of fog and different degrees of haze.

The invention discloses a ship maintenance progress prediction method based on multi-source data fusion and machine learning, and the method comprises the following steps: 1, obtaining multi-source data, and constructing a multi-source heterogeneous data set; 2, preprocessing and feature engineering processing are carried out on the multi-source heterogeneous data set, and the multi-source heterogeneous data set is converted into feature data needed by machine learning model input; 3, constructing a prediction model based on LightGBM and quantile regression, and training the prediction model through the feature data; 4, predicting the ship maintenance progress by using the trained prediction model; and 5, updating the prediction result of the ship maintenance progress in real time through Bayesian updating based on the progress performance index. According to the method, the project static attribute, the repair project network, the historical performance, the real-time resource state, the material supply information, the environmental factors and the streaming progress data are systematically integrated, the feature vector comprehensively describing the project state is formed, and the problems of complexity and uncertainty of the ship repair project are solved.

The utility model provides a friction coefficient detection device for thixotropic slurry, and relates to the technical field of pipe jacking engineering. The device comprises a moving device, the moving device comprises a pulley structure and a moving belt, the moving belt is arranged along the circumferential wall of a roller of the pulley structure to form a circulating moving system, a test sample structure and a standard test piece are arranged on the moving belt, the test sample structure is fixedly connected with the moving belt, and the standard test piece is arranged on the moving belt. The bottom of the standard test piece is in contact with the top of the test sample structure; the force measuring structure is connected with the standard test piece through a fixing device, when the pulley structure is moved, the moving belt and the test sample structure are driven to move, and friction is generated between the test sample structure and the standard test piece. According to the structure, the problems of inaccuracy and uncertainty of mud performance evaluation in engineering calculation and actual operation processes are solved.

This invention proposes a method for updating lane-level dynamic information in high-precision maps based on crowdsourced data. The method includes: collecting trajectory information and road image data from multiple sources, with a high-precision map embedded within; constructing a single-stage line-anchor detection model using a convolutional neural network; performing lane line perception detection on the road image data based on the single-stage line-anchor detection model; outputting lane line perception information; using an image alignment compensation algorithm to register and fuse road image data from consecutive frames during vehicle movement, and inputting this data into a target detection model for recognition, outputting traffic cone perception information; and projecting the lane line perception information and traffic cone perception information onto the high-precision map's geographic coordinate system using camera parameters. This method effectively suppresses missed and false detections in single-vehicle perception by combining low-cost crowdsourced data with a multi-vehicle evidence fusion mechanism, improving the accuracy and real-time performance of lane-level dynamic information updates, and meeting the high-precision requirements of autonomous driving for high-precision maps.

The invention discloses an approval process configuration method, system and device, a storage medium and a program product, and the method comprises the following steps: packaging a process engine into a WebAPI for terminal service calling through a workflow service module, and providing a BPMN-based workflow designer through the workflow service module; constructing an approval flow service module based on the workflow designer, defining an approval flow conforming to real businesses through the approval flow service module, and converting the approval flow into a BPMN definition; the examination and approval flow service module calls a flow engine through a WebAPI to analyze and process the BPMN definition so as to realize the operation of real business; wherein the terminal service comprises an approval flow service module. According to the method, visual definition and combination processes of various users can be met, and the demand of online circulation of actual business of an enterprise is met.

The invention provides a model and a method for predicting the water-gas ratio and the water yield of formation water of a fracture-vug type water-producing gas well. The model establishment method comprises the following steps: screening out abnormal fluctuation data from gas field production dynamic data to obtain daily output data; calculating the accumulated gas production and formation water-gas ratio of the water-producing gas well according to the daily output data; establishing a formation water-gas ratio calculation model of the water-producing gas well, wherein the model is a mathematical relationship between the formation water-gas ratio and cumulative gas production and contains a plurality of undetermined indexes; the accumulated gas production of the water-producing gas well and the water-gas ratio of the formation water are used for fitting to determine the value of each index, and then a prediction model of the water-gas ratio of the formation water is obtained. The water yield prediction model is established based on the formation water-gas ratio model. According to the prediction method, the future daily water yield can be predicted according to the water yield prediction model. According to the method, the labor cost of a large number of professionals is effectively reduced, the timeliness and convenience are improved, and the method has good applicability and popularization prospects in field practical application.

The application provides a high-robustness visual marker recognition method, system, device and medium for augmented reality surgery navigation, relates to the field of computer vision, and comprises the following steps: acquiring a surgery scene image collected by a collection device in surgery; detecting a candidate region of a marker in the surgery scene image based on a deep learning target detection network to obtain boundary box information of the marker region; accurately positioning a marker corner point in the marker region based on a sub-pixel corner point refinement network to obtain a refined corner point coordinate; and calculating a six-degree-of-freedom pose result of the marker corner point relative to the collection device based on the refined corner point coordinate and a predefined physical size of the marker corner point. The application overcomes the problems of insufficient robustness, limited corner point positioning accuracy and insufficient generalization ability of existing visual marker recognition technology in a high-dynamic imaging environment in an operating room, and realizes stable output of a pose result meeting the accuracy requirements of AR surgery navigation in a high-dynamic surgery environment.

The utility model discloses a kind of friction polygranulator grinding head positioning control mechanisms in the field of spandex staple waste silk processing, including shell, adjusting pivot is provided in shell, the end of adjusting pivot extends shell, the part of adjusting pivot extension shell is equipped with transmission wheel one, the top of shell is provided with rotary drive mechanism, the output of rotary drive mechanism is equipped with transmission wheel two, transmission wheel two is transmission connection with transmission wheel one by transmission part, mounting bracket is connected on the shell, encoder is installed on adjusting pivot corresponding mounting bracket, the extension end of adjusting pivot is transmission connection with encoder. The utility model uses the mode of encoder digital positioning to determine grinding head position, avoid the damage of equipment caused by too small grinding head gap, to reach the purpose of stable production.

The application discloses a multi-source micro-energy collection power distribution characteristic evaluation method and device and a storage medium, and relates to the technical field of micro-energy collection. Firstly, a multi-natural environment variable related coupling Copula function model is established according to historical wind speed, illumination, temperature and radio field intensity information of an application site, then a sampling point needing to be evaluated is established based on the probability density distribution characteristics of the natural environment, then the total power corresponding to the output of a mixed micro-energy collection system of each sampling point is calculated based on the power output characteristic curves of each micro-energy collection device, further, the multi-source mixed micro-energy collection system output power evaluation function under the sampling point is established with each natural environment variable as the dependent variable, the sampling point depth is adaptively increased by nesting until the final output power evaluation function is determined, finally, an output power sample sequence is generated, and the multi-source mixed micro-energy collection system output power probability density function is determined based on kernel density estimation.

This invention discloses a group train tracking operation optimization method considering communication delay, comprising the following steps: establishing a heavy-haul railway train group with a "leader-follower" topology; constructing a multi-objective optimization model based on vehicle characteristic constraints, speed limits, and safety distance constraints; generating random communication delays in the model, calculating the time steps corresponding to the delays, and having the following train make control decisions based on the state information of the preceding train at the previous time step; and solving for the running speed of the following train in the multi-objective optimization model using a particle swarm optimization algorithm. This group train tracking operation optimization method effectively solves the problems of uncertainty and dynamic response lag caused by communication delays during the tracking operation of heavy-haul railway train groups. It can not only significantly reduce tracking errors between trains and improve speed synchronization, but also effectively reduce distance fluctuations during acceleration and deceleration phases, ensuring stable train spacing.

The invention relates to the technical field of vehicle control, and discloses a formation vehicle parking method and device, and the method comprises the steps: detecting the information of a first available parking space in a driving process, and broadcasting the information of the first available parking space to a vehicle team, and the vehicles in the vehicle team are used for determining respective target parking spaces from the first parking space information capable of being parked according to the destination and driving to the first target parking space of the first vehicle. By applying the technical scheme of the invention, collaborative parking and information sharing among motorcades can be realized, and the parking efficiency is improved.