13results about How to "Improve description accuracy" patented technology

The invention belongs to the technical field of computer vision and artificial intelligence, and discloses a combined zero sample image classification method based on hierarchical feature fusion. According to the method, a solution based on hierarchical feature fusion is innovatively provided for the technical difficulty of combined zero sample image classification. According to the method, features of different depths in a CLIP model visual encoder are selected and subjected to hierarchical fusion, multi-level and multi-scale visual information is effectively extracted, the fused features are further used for visual and semantic cross-modal interaction, and the understanding ability of the model for complex combinations is remarkably improved. In addition, by introducing a specific loss item, attributes and object features are effectively separated, and the feature aliasing problem is avoided. In a combined zero sample image classification task, the method shows excellent generalization ability and recognition effect, and can significantly improve the classification performance of the model for unknown complex combined images.

The invention provides a vibration analysis method and system for an intermediate bearing in a double-rotor system. The method comprises the following steps: determining a rolling body entering a defect area at the current moment and a rolling body with the maximum contact deformation based on bearing parameters, an operation state, defect information, contact rigidity and damping parameters; respectively calculating the time-varying displacement and impact force of the rolling body entering the defect area and the rolling body with the maximum contact deformation, and the angle difference between the time-varying displacement and the impact force; the impact force is corrected through the angle difference between the rolling body entering the defect area and the rolling body with the largest contact deformation and serves as external force input of a nonlinear kinetic equation of an intermediate bearing system of the double-rotor system; and calculating a nonlinear bearing force in the nonlinear kinetic equation based on the time-varying displacement of the rolling body entering the defect area and the rolling body with the maximum contact deformation. And solving the nonlinear kinetic equation. According to the method, the vibration analysis precision under the condition of the defect of the intermediate bearing in the double-rotor system is improved.

The present application relates to a method, device, design method and device for optimizing a force field and its non-bonded parameters. The method for optimizing the non-bonded parameters of the force field includes: obtaining M clusters; determining the quantitative intermolecular interaction energy of each polymer of at least some of the M clusters; for each cluster in the M clusters, fitting the electrostatic potential of the central molecule of the cluster based on the background charge of the cluster to obtain the atomic charge of each atom in the central molecule, and then obtaining the charge parameters of the force field based on the atomic charge of each atom in the central molecule of the M clusters; based on the quantitative intermolecular interaction energy of each polymer of the clusters, using the atomic charge of each atom in the central molecule as the initial charge of each atom of the molecule in the polymer, fixing the bonding parameters and the initial charge unchanged, and fitting to obtain the van der Waals parameters of the molecule. The present application can improve the accuracy of the description of polarization by force field parameters.

The invention is applied to the technical field of data processing, particularly can be applied to various scenes such as cloud technology, artificial intelligence, intelligent traffic and auxiliary driving, and relates to an information processing method and device, electronic equipment and a storage medium, and the method comprises the steps: building graph data based on the historical operation conditions of each target object for a business object in a specified business scene; the graph data comprises a node information set constructed based on nodes corresponding to each target object and each business object, a time information set constructed by operation time corresponding to a connection edge, node features and edge interaction features; and based on the graph data, performing multi-round iterative training on the initial graph neural network model to obtain a trained target model and target embedding vectors constructed by the target model for each node. Therefore, the business object range which can be considered by the target object is improved, and a favorable analysis basis is provided for analyzing the adaptability of the business object to the target object.

A twist drill device for detecting the compaction of foundation soil includes a column connected to the drilling rig's transmission. The column is connected to the twist drill via a torque ring, and multiple blades are installed on the twist drill. The torque ring includes a ring body with an internal thread at the top and an external thread at the bottom. A torque sensor and a cavity are installed within the ring body. A Bluetooth module, a power supply, and a microprocessor are housed within the cavity. The torque sensor is connected to the microprocessor, which is connected to a remote smart terminal via the Bluetooth module. This invention directly installs a torque ring device between the drilling rig column and the twist drill. This device can measure the torque and pressure exerted by the column on the twist drill in real time and transmit the torque and pressure data to a smart terminal via Bluetooth technology. This invention solves the problem that core samples obtained using twist drills are in a completely disturbed state, making it impossible for the operator to determine the compaction of the foundation soil on-site.

The intelligent classification method for artificial building scenes in high-resolution remote sensing images of this application establishes a vision-oriented hidden layer probabilistic semantic analysis scene classification method, adds spatial information constraints to the features of scene expression, carries more semantic information with lower feature dimensions, and improves the expressive power of visual vocabulary. First, the image performance characteristics of typical artificial scenes are analyzed in combination with the data features of remote sensing images, and a lower feature dimension is used to describe the target visual information; second, the hidden layer probabilistic semantic analysis model is improved in combination with the features of remote sensing images, and the strategy of segmenting the most characteristic vocabulary of vision is adopted to describe the visual features of the over-segmented image, and the rich image spatial information and geometric structure information brought by high spatial resolution are used to make up for the limited description power of local feature points; by combining the vision-oriented classification method with the focus model, a better classification effect is achieved with low-dimensional feature description.

The invention relates to the field of data processing, in particular to a multivariate big data optimization storage method, which comprises the following steps of: obtaining a plurality of multivariate data points; obtaining a comparison point, and calculating a correlation description value of each dimension of the comparison point and other multivariate data points; performing clustering processing on all the multivariate data points according to the correlation description values; acquiring a first principal component direction, and acquiring a reference data point according to the first principal component direction; obtaining a unit vector and a modulus length of a description vector of each multivariate data point according to the reference data point; and taking the unit vectors as coding objects, performing Huffman coding on the unit vectors of all the multivariate data points in the category, and storing codes and die lengths corresponding to the unit vectors of the multivariate data points. The data compression amount is improved by selecting proper coding features.

The invention discloses a tight constraint model correction method based on hybrid functional and defect calibration. According to the method, an initial energy band of a defect-containing supercell is obtained based on a standard PBE functional, and energy band correction is carried out by utilizing a band gap difference value of a defect-free reference unit cell between the PBE functional and an HSE functional. In combination with charge transition energy level and energy window constraint of HSE functional calculation, an extrinsic defect energy band is analyzed and deentangled through wave function locality or projection energy band, the extrinsic defect energy band is integrally translated to transition energy level alignment, and a high-precision calibration energy band is obtained. The initial energy band and the calibration energy band are sequentially put into a graph neural network for preliminary training and continuous training so as to accurately extract tightly bound model parameters. And constructing a device channel Hamiltonian by means of the parameter, and executing quantum transport simulation in cooperation with a non-equilibrium Green Function (NEGF). According to the method, synchronous high-precision correction of the band gap and the defect energy level is realized with relatively low calculation cost, and a reliable scheme is provided for device-level quantum transport simulation of a defect-containing semiconductor system.

The invention relates to the technical field of radar echo modeling, in particular to a radar clutter distribution trailing modeling method, a radar clutter distribution trailing characteristic analysis method and a radar clutter distribution trailing characteristic analysis device, and aims to construct a polynomial model of radar clutter distribution trailing by performing polynomial fitting on a probability density function of clutter distribution trailing so as to accurately describe the trailing characteristics of radar clutter distribution. The structure is simple, and engineering realization is easy; actual measurement data verify that the method has a better fitting effect, and the accuracy of clutter distribution trailing characteristic description can be effectively improved.

This invention relates to the field of mechanical transmission and friction and wear analysis, and discloses a method, system, medium, and equipment for constructing a dynamic wear model of a wedge-type one-way clutch. The method includes: establishing an analytical model of the dynamic interface force of the wedge, solving for the force and contact area; obtaining the contact pressure distribution, relative sliding velocity, and contact time of the wedge under different working conditions; constructing a wear volume evolution model over time; introducing a wear constitutive equation to establish a quantitative relationship between contact pressure, relative sliding distance, and wear amount; integrating and coupling the wear volume evolution model, the wear constitutive equation, and the dynamic interface force analytical model, correcting the contact friction coefficient and wear coefficient by temperature field factors, and constructing a multi-physics coupling framework; dynamically updating the contact surface geometry based on the wear accumulation results, and recalculating the contact force and wear parameters to form a continuous iterative wear evolution process. This invention can achieve dynamic and continuous prediction of wedge wear, improving the accuracy of wear assessment under complex working conditions.

The present invention relates to the technical field of defect recognition, and in particular to a method and system for identifying defects in clothing fabrics, the method comprising: collecting a grayscale image of clothing fabrics; obtaining scale-invariant feature points in the grayscale image of clothing fabrics; obtaining the main direction of each single structure image in the edge binary image of the clothing fabrics grayscale image; determining the feature descriptors of the scale-invariant feature points in each dimension according to each edge point in the pixel window of the scale-invariant feature point and the main direction of the single structure image; using the SIFT algorithm to complete image matching based on the feature descriptors of the scale-invariant feature points, performing defect recognition according to the matching results, and judging the quality of clothing fabrics. The present invention aims to mine the pattern features of clothing fabrics and improve the accuracy of defect recognition.

This invention discloses a stability determination method and system for an automatic voltage control closed-loop system in a new energy base. The method includes: acquiring system parameters of the automatic voltage control closed-loop system; establishing a discrete state update model including the current cycle control command and the previous cycle control command; constructing augmented state variables and establishing a unified closed-loop discrete mapping relationship considering control laws and limiting constraints; dividing the state space into an unsaturated subspace and a saturated subspace according to whether the control command touches the limiting constraint, and establishing corresponding piecewise discrete mapping relationships; constructing a linear matrix inequality stability criterion for the unsaturated subspace under the condition that the voltage-reactive power characteristics satisfy the bounded incremental sector, and transforming it into a semi-positive definite programming problem to obtain the stability determination result. This invention can accurately characterize the dynamic behavior of the AVC closed loop in a new energy base under the combined effects of sample-and-hold, time delay, and limiting, and realize the analysis and determination of the system's stability state and oscillation mode.

The application provides an infrared-optical fusion scene semantic description method and device based on a large model, and belongs to the technical field of image description and computer vision. The method comprises the following steps: inputting a visible light image pair and an infrared image pair obtained to a preset scene description model to obtain an initial scene description sentence of the image pair; inputting the infrared image of the image pair to a preset target detection model to obtain the category of each target detection frame corresponding target in the infrared image, and the center point coordinates and relative size of the target; inputting the above information to a large language model with prompt optimization completed, and the large language model outputs a final scene description sentence corresponding to the image pair. The target detection technology and the scene description technology are combined through the large language model technology, the defects that the previous image description model cannot be applied to a dark light scene or is not accurate in description can be overcome, and especially, the scene description can be generated accurately and in detail under insufficient light, so that the application value is high.