10results about How to "Fast results" patented technology

This invention provides a data processing method and system based on an AI chip, relating to the field of electronic digital data processing technology. The method includes: acquiring the current data unit in a target data stream; calculating its data similarity C with previous data units; and calculating the entropy decay index D between adjacent computational layers in the layer-by-layer processing of a neural network. {N} The entropy decay index D {N} Used to quantify the degree of decrease in information uncertainty; the data similarity C and entropy decay index D {N} Perform a joint judgment; when C is higher than the similarity threshold and D... {N} When the data falls below the decay threshold, processing of subsequent computational layers is interrupted, and the feature data of the current layer is discarded as redundant data. This invention achieves dynamic pruning of the computational path by constructing a dual verification criterion of input similarity and entropy change in the model's internal processing. This reduces redundant computation and improves the processing efficiency and energy efficiency of AI chips without sacrificing risk identification accuracy.

This invention belongs to the field of aircraft design and strength analysis technology, specifically relating to a rapid design and optimization method for openings in aircraft structures. The method involves: establishing a full-aircraft model for overall strength analysis; determining the location and size of the required openings on the finite element model; selecting a specific finite element region for analysis; analyzing the severe operating conditions of the finite element elements in this region; outputting the displacement and stress at the boundaries of the selected elements under severe operating conditions; importing the model and boundary conditions into software with a self-developed opening design and analysis module; performing optimized opening design and strength analysis; forming a reasonable opening reinforcement scheme; and outputting the calculation results. The method proposed in this invention can reduce weight, decrease local stress concentration, and delay or prevent the occurrence of cracks at the hole edge.

A method and system for image similarity recognition based on multi-feature fusion, relating to the field of image processing technology, is disclosed. The method includes: calculating the hash features of an input image and generating hash similarity and hash confidence values; when the hash confidence value is less than a hash confidence reference value, performing local feature extraction on the input image to generate local similarity and local confidence values; when the local confidence value is less than the local confidence reference value, performing deep learning feature extraction on the input image to obtain deep features and deep confidence values; determining the weight coefficients of the hash features, local features, and deep features based on the hash confidence value, local confidence value, and deep confidence value, respectively; performing feature fusion to generate a fused feature vector; and calculating the final similarity between the input image and the image to be compared based on the fused feature vector. Implementing this application can improve processing efficiency while ensuring the accuracy of image similarity recognition.

This invention provides a finite element method for solving generalized axisymmetric plane problems using stress as the fundamental variable, relating to the field of engineering analysis. The method includes: S1. Selecting basis functions that satisfy the equilibrium equations; S2. Applying the complementary energy principle to establish the element equations for each element; S3. Assembling all element equations to form the overall stiffness equation; S4. Applying forced boundary conditions to the stiffness equation; S5. Solving the stiffness equation; S6. Obtaining all stress components at any point within the element through basis function interpolation; S7. Repeating S1 to S6 to obtain the element stress distribution under each basic load condition; S8. Obtaining the stress state under any composite load based on the principle of linear superposition; S9. Evaluating the strength of the stressed body according to material failure criteria. This invention does not require users to possess professional modeling skills; through simple parameter input, the stress solution of complex axisymmetric problems can be obtained quickly and accurately, and strength evaluation and performance characterization can be performed accordingly.

The invention provides a finite element method for solving a generalized axisymmetric plane problem by taking stress as a basic variable, and relates to the field of engineering analysis, and the finite element method comprises the following steps: S1, selecting a primary function meeting an equilibrium equation; s2, establishing a unit equation of each unit by applying a complementary energy principle; s3, assembling all unit equations to form an overall rigidity equation; s4, applying a forced boundary condition to the stiffness equation; s5, solving a stiffness equation; s6, obtaining all stress components of any point in the unit through primary function interpolation; s7, repeating the steps S1 to S6, and solving to obtain unit stress distribution under each basic load working condition; s8, based on a linear superposition principle, obtaining a stress state under any composite load; and S9, performing strength evaluation on the stress body according to a material failure criterion. According to the method, a user does not need to have professional modeling skills, the stress solution of the complex axial symmetry problem can be quickly and accurately obtained through simple parameter input, and strength evaluation and performance characterization are carried out according to the stress solution.

The invention provides an alpha energy spectrum quantitative method, device and equipment based on a multi-population particle swarm, and a medium, and the method comprises the steps: enumerating the thickness of a sample, and calculating a response function matrix corresponding to the samples with different thicknesses; enabling each thickness sample to correspond to one population, and establishing a relation equation among detector counting, nuclide activity and a response function matrix of each population; parallel energy spectrum analysis calculation is carried out on all the populations, particles in each population are iteratively searched through a speed updating formula and a position updating formula to obtain corresponding nuclide activity and energy spectrum fitting degree optimal solutions, in the parallel calculation process, the populations which cannot be converged are eliminated, the optimal population is screened out from the remaining populations, and the optimal solution of the nuclide activity and the energy spectrum fitting degree is obtained. The thickness corresponding to the optimal population is the optimal solution of the sample thickness, and the global optimal position corresponding to the optimal population is the optimal solution of the nuclide activity, so that the problems of complex sample pretreatment, long measurement period and incapability of directly analyzing the sample with the unknown thickness in the alpha energy spectrum measurement technology are solved.

The invention relates to the technical field of road detection and artificial intelligence, in particular to an unmanned aerial vehicle highway disease intelligent detection system based on deep learning. The system comprises an unmanned aerial vehicle platform, a data acquisition module, an image analysis and processing module, a deep learning disease detection module, a result output and early warning module and a ground work station, wherein the image analysis and processing module, the deep learning disease detection module and the result output and early warning module are all arranged in the ground work station; the data acquisition module is carried on the unmanned aerial vehicle platform, and the unmanned aerial vehicle platform carries a high-definition camera and is used for flying over an expressway according to a preset route. According to the invention, by integrating the unmanned aerial vehicle inspection platform and the YOLOv8-based deep learning model, a set of full-process automatic detection system integrating efficient data acquisition, precise disease identification and intelligent early warning decision is constructed, and the efficiency, precision and intelligent level of highway disease detection are significantly improved.

The utility model discloses a kind of liquid refractive index measuring devices, belong to measuring device technical field, including bottom plate, the bottom plate top is symmetrically provided with slide rail, the slide rail top is from left to right array and is provided with four sliding bases, the sliding base top is fixedly installed with support, positioning mechanism for positioning sliding base is arranged on the sliding base, four sliding bases are sequentially fixedly connected with laser light source, beam expander, adjustable single slit, double slit plate and light source receiver by support from left to right, two light slots are provided on the double slit plate, colorimetric dish and a wedge glass are fixedly connected on the double slit plate side;The utility model uses screw micrometer to read numerical value, replaces traditional guide rail reading, and self-made controllable optical path device. By adjusting the thickness of the compensation device, the refractive index measurement range is increased, the movement of interference fringes is more intuitive, thereby the flexibility of liquid refractive index measuring device is improved.

Systems and methods for determining the concentration of a plant component in a developed food product prior to or during the manufacture of the food product are disclosed. The method may include: receiving, by a computing system, protein data of a plant component; retrieving at least one model, the at least one model having been trained to generate an output indicative of a desired concentration of a plant component to be used for developing the food product, the at least one model comprising at least one of a pH model, a temperature model, and a protein content model; providing the received protein data as input to the model; receiving an output from the model; and generating instructions for developing the food product, the instructions including a desired concentration range of the plant ingredient.