177 results about "Feature structure" patented technology

The invention discloses a method for controlling the machining accuracy of large integrated thin-walled parts based on finite element analysis. The method comprises the following steps of: 1, extracting local structural features of the large integrated thin-walled part; 2, performing finite element simulative analysis on local structural features to obtain an optimized cutting technology; 3, numerically modeling a large integrated workblank, and loading an initial internal stress; 4, protocoling a tool path, namely protocoling the machining sequence of the feature structures; 5, performing simulative analysis on the integrated structure to obtain a predicted deformation result under the condition of process technology; 6, regulating and optimizing a clamp scheme to control machining deformation. The method adopts finite element simulation, can analyze and predict in advance before the actual machining of the parts, and thereby corresponding machining strategy is adjusted, machining deformation is effectively controlled, production cycle of the parts is shortened, and production cost is reduced.

A method and apparatus is featured for providing linear control of a regulated electrical property (e.g., current or voltage) in a switch mode power supply of a thermal processing system that includes an inductive element and at least one switching element. The method and apparatus feature structure, or steps, for generating a nonlinear model that predict values for a regulated electrical property of the inductive element based on a given duty cycle of the at least one switching element and structure, or steps, for generating a nonlinear model that determines a duty cycle for the at least one switching element based on the nonlinear predictive model for the regulated electrical property, the nonlinear model determining the duty cycle such that a linear relationship results between the regulated electrical property and a selected value. A method and apparatus is also featured for providing cross regulation between pre-regulator and post-regulator power stages of a switch mode power supply of a thermal processing system.

The invention belongs to the technical field of nano materials, and in particular relates to a method for preparing a hybrid composite material of carbon fibers / carbon nanotubes / bismaleimide resin. The method comprises the following steps: after carboxylating, acylating and chlorinating the surfaces of purified carbon nanotubes and dried carbon fibers, introducing diamine or polyamine with a feature structure thereon, wherein a large amount of the diamine or the polyamine are jointed on the surfaces of the carbon nanotubes and the carbon fibers; performing an addition reaction on the carbon nanotubes jointed with the amidogen and the bismaleimide resin to obtain a carbon nanotubes-containing bismaleimide resin linear block polymer serving as a substrate; and compositing the substrate and the carbon fibers in a certain mode to form a multi-dimensional hybrid composite material structure linked by covalent bonds finally. The method has the advantages of strengthening and toughening bismaleimide by using the strength and the toughness of the carbon nanotubes, improving the bonding strength between the bismaleimide and the carbon fiber substrate surface, improving overall performance of the multi-dimensional hybrid composite material of the carbon fibers / the carbon nanotubes / the bismaleimide resin, and widening the application of the carbon fibers, the carbon nanotubes and the bismaleimide resin.

The invention relates to the field of ground object classification, in particular to a ground object classification method and device based on a multispectral image and an SAR image, and the method comprises the steps: obtaining the multispectral image of a preset area, and carrying out the multispectral image feature extraction of the multispectral image; obtaining a time sequence SAR image of apreset area, and performing time sequence SAR image feature extraction on the time sequence SAR image; and performing feature level fusion on the multispectral image features and the time sequence SARimage features to obtain a ground object classification result. According to the method and the device, the advantages of all-day working, all-weather working and short revisit period of the synthetic aperture radar SAR are utilized to obtain a long-time sequence SAR image, and the input feature dimension is increased; Characteristic level fusion is carried out on multispectral images and SAR images, and while multispectral information is fully utilized, ground feature interpretation is assisted by combining ground feature structures, textures and electromagnetic scattering characteristics reflected by time sequence SAR images.

The invention discloses a metal plate forming device which comprises a mold and an electromagnetic production device, wherein a forming mask of the mold has a pattern structure, and the pattern structure comprises at least one high part and at least one low part; a plate-shaped electromagnetic coil of the electromagnetic production device has a forming region, and a feature structure corresponding to the pattern structure is arranged in the forming region; and a metal plate can be driven by a mutually-repelled electromagnetic force generated between the plate-shaped electromagnetic coil and the metal plate to abut on the pattern structure, thereby forming the preset pattern. Thus, the metal plate forming device has the following advantages: the lateral binding force of the blank (metal plate) in the forming process can be reduced; the formed pattern has a higher depth-width ratio; the forming performance of the blank is improved; the thinning rate and warpage problem of the blank can be reduced; the structure is simple; a higher primary current value can be produced; and the arcing discharge problem can be completely avoided.

The invention relates to a text watermark embedding and extracting method based on character structure characteristics, which comprises the following steps: 1) according to the used word frequency, building a character object set, modifying characteristic points of the character objects to obtain a new work stock file; 2) installing the new work stock file at a computer terminal and printing the text, when printing and outputting the text, intercepting and capturing the text content data and analyzing the character objects, dynamically substituting watermark information to the character objects in the character object set; and 3) obtaining digital image data of the printed text by an image copying device, analyzing characteristic point information of each character designated location in the text image, and extracting the watermark information. The quantity of the carrier objects of the embedded watermark can be a lot; and the extracting precision of the watermark information is relatively high. At the same time, the modification and analysis principles of the character structure characteristic information are simple and clear; the character structure characteristic information is normalized, so that zoom and attack at unequal resolution ratio can be resisted; and the robustness is relatively high.

The invention discloses a method for automatically identifying and distinguishing eye fundus images. The method comprises the following steps: obtaining black and white or colored eye fundus image pictures by using eye fundus photographic equipment, and storing the black and white or colored eye fundus image pictures according to a left eye or a right eye; dividing the eye fundus image of the left eye or the right eye into seven regions, which are respectively a first region-optic disk region, a second region-macular region, a third region- macular temporal region, a fourth region- area temporalis superior, a fifth region- area temporalis inferior, a sixth region- superior nasal region and a seventh region- inferior nasal region; and automatically judging that the collected eye fundus image belongs to one of the seven eye fundus regions through the computer graphics according to the optic disk, macular and vascular network information in the eye fundus image. The method disclosed by the invention is used for introducing automatic computer image identification into the processing of the images of the seven eye fundus regions, and extracting feature structures of different eye fundus regions to serve as reference for automatically locating and collecting the eye fundus image, and providing a brand new manner for researchers or film reading doctors to research and read the images of the seven eye fundus regions, so as to greatly improve the film reading efficiency and reduce the error probability.

The invention discloses a laser-based inside-laser coaxial wire-feeding additive manufacturing system and a forming method. The laser-based inside-laser coaxial wire-feeding additive manufacturing system is formed by a laser, a wire feeder mechanism, a movement actuating mechanism, a molten pool shape monitoring feedback module and a system control module. After a part model is cut into slices, aninitial path is planned, input parameters and technological parameters in a technology database are compared, the molten pool temperature corresponding to technological parameters and input technology parameters under different feature structures is optimized, printing parameters and a robot code under an optimal path are generated, a high-speed infrared camera acquires molten pool shape parameters during the printing process and transmits to the system control module, the system control module compares the acquired molten pool shape data and the technology data of the technology database module, meanwhile, the molten pool temperature change is monitored, and the molten pool temperature is corrected through feedback-adjusting laser power and is monitored and feedback-adjusted in real time, so that the molten pool shape is uniform and consistent and the printing process is stably controlled during the printing process.

The invention discloses a low-dosage CT image decomposition method based on three-dimensional distinctive feature representation. The method comprises steps of: scanning a phantom to acquire a group of corresponding low-dosage and normal-dosage phantom CT images; selecting a feature block in the normal-dosage phantom CT image to form a feature dictionary, performing subtraction on the low-dosage and normal-dosage phantom CT images so as to obtain a low-dosage noise pseudo shadow image, and selecting a feature block in the noise pseudo shadow image to form a noise pseudo shadow dictionary; and representing a clinic low-dosage CT image by using a three-dimensional distinctive dictionary formed by the feature dictionary and the noise pseudo shadow dictionary in order to obtain a feature image represented by the feature dictionary and a noise pseudo shadow image represented by the noise pseudo shadow dictionary, thereby achieving decomposition of the low-dosage CT image. The method may effectively separate noise and strip-shaped pseudo shadow from feature structure components in the low-dosage CT image, satisfies a quality requirement of clinic analysis and diagnosis, and improves the use efficiency of the low-dosage CT image.

The invention belongs to the technical field of transformer detection, and discloses a YOLO network model-based transformer component identification method, which comprises the following steps: acquiring a monitoring image, and carrying out denoising and size scaling preprocessing; calibrating components of the transformer in the image; carrying out clustering on the data set to generate an appropriate priori frame scale; inputting the calibrated image into a deep neural network model for training and continuously adjusting parameters, so that the network model can automatically detect and position main components of the transformer, has a classification capability, and identifies which class the main components of the transformer belong to; carrying out detection positioning and state recognition by using the trained model; According to the invention, main parts of the transformer equipment in the transformer substation monitoring image can be automatically detected and identified; compared with the prior art, the method has the advantages that the method is high in recognition accuracy, the model uses a fine-grained feature structure, that is, depth features are fused, differentfeature maps adapt to targets of different scales, and a good recognition effect on small target detection is achieved.

It is an object of the present invention to provide a method and apparatus for locating a keyword of a speech and a speech recognition system. The method includes the steps of: by extracting feature parameters from frames constituting the recognition target speech, forming a feature parameter vector sequence that represents the recognition target speech; by normalizing of the feature parameter vector sequence with use of a codebook containing a plurality of codebook vectors, obtaining a feature trace of the recognition target speech in a vector space; and specifying the position of a keyword by matching prestored keyword template traces with the feature trace. According to the present invention, a keyword template trace and a feature space trace of a target speech are drawn in accordance with an identical codebook. This causes resampling to be unnecessary in performing linear movement matching of speech wave frames having similar phonological feature structures. This makes it possible to improve the speed of location and recognition while ensuring the precision of recognition.

The invention provides a point-cloud feature point extraction method based on local sub-neighborhood division. The method includes the four steps of the initial feature point extraction stage, the local triangle construction stage, the local sub-neighborhood division stage and the robust feature point extraction stage. At the initial feature point extraction stage, potential initial feature points of input point cloud data are acquired. At the local triangle construction stage, in order to effectively extract true feature points from an initial feature point set, a triangle set capable of reflecting the local geometrical feature structure of the corresponding point is established in a local neighborhood of each initial feature point. At the local sub-neighborhood division stage, the normal of each constructed local triangle set is clustered, local neighborhood points at one point can be clustered, and thus sub-neighborhood division of the local neighborhood points at one point can be achieved. At the robust feature point extraction stage, local plane fitting is conducted on the data points in the divided sub-neighborhood, and accordingly the true feature points can be recognized by judging whether current points fall on intersection lines of a plurality of planes at the same time or not.

The invention relates to a PDMS (polydimethylsiloxane)-based functional polymer patterning method. The method comprises the steps of: firstly, obtaining the required micro-nano size of a pattern at the surface of a light-sensitive lacquer by using a lithography method; coating an un-crosslinked organic material PDMS at the surface of the patterned light-sensitive lacquer, and taking down the organic material PDMS after cross-linking, so as to finish the fabrication of a soft template; spinning one surface of the soft temperature, provided with the structure, on a functional polymer substrate; and achieving the patterning of different functional polymers by the hot stamping or the ultraviolet imprinting. A hard template does not need to be defined as an initial structure, and just the patterning of the light-sensitive lacquer is needed. Meanwhile, the anti-sticking treatment is not needed in a PMMA (polymethyl methacrylate) template for the initial definition; the pattern of the PMMA can be transferred; the experimental procedure can be reduced; the transfer cost is reduced; and the fidelity of the pattern is improved. Furthermore, the PDMS soft template does not need the anti-sticking treatment. Meanwhile, the requirement on the levelness of a final substrate B is low; and the imprinting of a large-range feature structure can be achieved.

The invention discloses a locality preserving PCA-based three-dimensional point cloud registration method. In order to preserve local features of a point cloud, an LPP (Locality Preserving Projection)thought is adopted, and an adjacency graph and a complement graph of the point cloud are constructed through a K-neighbor criterion; adjacent points and non adjacent points are subjected to feature extraction in different processing modes; a conversion parameter is calculated through a feature matrix; and coordinate normalization is performed to finish registration. By adopting the technical scheme, a relatively good effect is achieved in registration of the point cloud with a remarkable local feature structure.

The invention provides a springback compensation method for creep age forming. The method comprises the following steps: Step S100, based on MSC.MARC commercial finite element software, constructing creep age forming finite element simulation; Step S200, weighting a springback iterative compensation process based on a molding surface of a finite element simulation result; and Step S300, considering a reverse compensation process of high temperature molding surface thermal expansion effect. The method considers the influence of the strength of a component on surface compensation, the influenceof a component feature structure on surface compensation, and reverse compensation of thermal expansion of a high temperature molding surface. A component formed by the model can still meet the engineering error requirement after being cooled.

The invention discloses a multi-scale structural material integrated design method based on shape features. The method is used for solving the technical problem that existing structural material integrated design methods are poor in practicability. According to the technical scheme, the shape features are used as basic elements of macro and micro design, a regular quadrangular finite element gridis adopted, and a structural shape is described by calculating a level set function of a feature structure; material properties of a microstructure are associated with macrostructure units through a homogenization method, and a multi-scale mechanical model based on a fixed grid is established; and shape position parameters of the shape features are selected to serve as design variables, and macroand micro topological optimization design is realized. Through the method, macrostructure design and microstructure design are associated, the level set method is adopted to describe the model, multiple microstructure forms can be obtained according to local conditions, an optimization result which has a smooth boundary and apparently contains feature information can be obtained, the method can beseamlessly integrated with existing computer-aided design software, and practicability is high.

This invention relates to ionized PVD processing of semiconductor wafers and provides conditions for highly uniform deposition-etch process sequence and coverage capabilities of high aspect ratio (HAR) features within a single processing chamber. A plasma is generated and maintained by an inductively coupled plasma (ICP) source. A deposition process step is performed in which metal vapor is produced from a target of a PVD source. Location and sputter efficiency at the target surface is enhanced by moving a magnet pack to create a traveling or sweeping magnetic field envelope. The target is energized from a DC power supply and pressures effective for an efficient thermalization of the sputtered atoms (30<p<100 mTorr) are maintained in the chamber during deposition. A uniform thickness of the metal on the wafer is produced within each magnet sweeping cycle. Magnetic field localization using an annular sweeping motion over the entire target surface generates conditions for reasonable deposition rates, high target utilization, high ionization of the metal atoms, uniform flat field deposition and etching, and nearly identical conditions for HAR feature coverage at the center and edge of the wafer.

Embodiments described herein generally relate to an apparatus and method for uniform sputter deposition of material onto the bottom and sidewalls of high aspect ratio features on a substrate. In one embodiment, a collimator positioned between a sputter target and a substrate support for mechanical and electrical connection with a shield member is provided. The collimator includes a central region and a peripheral region, wherein the collimator has a plurality of apertures extending therethrough, and wherein the apertures in the central region have a higher aspect ratio than the apertures in the peripheral region.

According to the invention, a super-resolution reconstructed picture is more consistent with a human eye visual perception effect; at the same time, the feature structure of the picture is maintained.The invention provides an image super-resolution reconstruction method based on no-reference quality evaluation and feature statistics. An adversarial learning network model is constructed, more high-frequency detail edges are generated, the image looks like more textures, the tone, the brightness and the sharpness of the image are adjusted through no-reference quality evaluation, so that the generated image is closer to human visual perception, and the internal feature structure of the image is maintained through feature statistics. Compared with a traditional super-resolution reconstructionmethod, the high-resolution image generated through the method has more abundant real texture details, the human eye perception effect of the image is improved, and the result content of the image cannot be damaged.

The invention discloses a semantic segmentation improved algorithm based on a point cloud local structure, which is used for assisting the dense mapping and belongs to the field of multimedia signal processing. The semantic segmentation improved algorithm comprises the following steps of 1, learning a local geometric structure; 2, learning a local feature structure; 3 fusing the ORB-SLAM2 systems.The present invention provides the algorithm for executing the semantic segmentation service based on a kernel association network for the first time, the algorithm enables the accuracy to be improved to 80.6% under an S3DIS data set and assists the ORB-SLAM2 with a dense point cloud mapping function, so that the SLAM system has the semantic mapping function.

The invention discloses a real-time phased array signal distortion correction method based on interference suppression. Distortion of estimation on a steering vector of an unknown radar target echo in actual engineering is the result of an amplitude-phase error between channels of a phased array receiving antenna, and the output of the radar target echo can be distorted. According to the real-time phased array signal distortion correction method disclosed by the invention, the steering vector of the radar target echo is precisely estimated, and an amplitude and phase difference between two adjacent weight vector updating moments of a phased array output signal is obtained in real time; therefore, real-time and effective correction is implemented according to the difference; according to the two adjacent weight vector updating moments, if the signal energy corresponding to the later updating moment is higher, secondary correction needs to be executed. The real-time phased array signal distortion correction method can be implemented on any feature-structure-based direction estimation array antenna; the effect of real-time and precision correction for the distortion of a target output signal, which is caused by the amplitude-phase error and change interference between the channels of the receiving antenna, is achieved under a low signal to noise ratio. Theoretical analysis and an experiment result show that a real-time, precision and stable correction effect can be achieved, and the real-time phased array signal distortion correction method is suitable for actual engineering practice.

The invention provides a method for preparing acrylamide complete antigen. The method is as follows: amine uncoupling reaction is directly carried out between N-acryloyloxy succinimide and the surface amino group of carrier protein molecule in a neutral or weakly alkaline phosphate buffer solution, thereby forming acrylamide complete antigen. Polyclonal antibody capable of realizing specificity identification of acrylamide is obtained through a complete antigen immune animal, so as to establish an enzyme immunoassay method suitable for measuring the content of acrylamide in an aqueous solution or food and to provide a corresponding reagent box. The method has simple operation and does not need other reaction reagent, thereby furthest maintaining the structural integrity of acrylamide; moreover, the method ensures that a coupling position is at a maximum distance from a feature group so as to fully expose the feature structure of acrylamide molecule. The established enzyme immunoassay method of acrylamide and the reagent box have strong specificity, high sensitivity and convenient operation, and are suitable to realize massive sample detection which has low cost and is quick and simple.

The invention discloses a method and a system for automatically detecting an academic misconduct literature. The method comprises the following steps of: establishing characteristics of a hierarchical content of a literature to be detected and table data in the literature; establishing characteristics of hierarchical contents of stored literatures and table data in the literatures; matching the hierarchical content characteristics of the literature to be detected and the table data characteristics in the literature to be detected with the hierarchical content characteristics of all the stored literatures and the table characteristics in the stored literatures; and judging whether the literature to be detected contains academic misconduct contents , academic misconduct table data and types of the academic misconduct contents or not. The system comprises a characteristic region of the literature to be detected, a comparison resource region of the literature to be detected, a hierarchical content characteristic region and a misconduct academic literature and type judgement region. Through a hierarchical multi-stage characteristic structure, the invention can fast detect an ultra-long literature, satisfy the detection of a minimum characteristic granularity short verse of the literature and enhance the pertiency factor and the recall ratio. In addition, the invention also supports to establish the table data characteristics in the literature and disposable matching for matching all the literatures.

The invention discloses an eyeball tumor area real-time positioning method based on motion features of an eyeball. The eyeball tumor area real-time positioning method comprises the following steps of extracting an eye anterior-segment tissue information, conducting calibration and three-dimensional reconstruction on the eye anterior-segment tissue information to obtain shifting parameters before and after eyeball motion, establishing a three-dimensional eyeball model based on a CT image, and establishing the coordinate relation between an eye anterior segment feature structure and an intraocular tumor position. According to the eyeball tumor area real-time positioning method, an anterior-segment image obtained through real-time OCT tracking is utilized, and then the tumor is accurately positioned in a radiation area through processing technologies including image reconstruction, registering and the like. The invention further discloses an imaging device for extracting the eye anterior-segment tissue information. The imaging device for extracting the eye anterior-segment tissue information comprises an OCT light source, a beam splitter, a sample arm, a reference arm and a feedback device. A real-time positioning tracking system provides in-vitro noninvasive positioning and can achieve the purposes of improving positioning accuracy and radiation dose and local control rate of the ocular tumor, decreasing surrounding normal radiation dose and injury and the like.

The present invention generally provides an apparatus and a method for inspecting a substrate in a substrate processing system. In one aspect, a voltage or current source is used in conjunction with a power density receiving device, such as a spectrometer, to inspect a substrate for various noise spectrum signatures. In one embodiment, spectral data collected from a given substrate is used to generate a current or voltage spectral signature. This spectral signature may then be compared to a reference spectral density signature to predict reliability of a feature structure of a substrate in processing and feedback to the substrate processing system for substrate processing control. Embodiments of the invention further include computer-readable media containing instructions for controlling the substrate processing system, and computer program products having computer-readable program code embodied therein for controlling the substrate processing system and inspecting defects on semiconductor features.

Methods and apparatus for depositing a metal-containing layer on a substrate are provided herein. In some embodiments, a method of processing a substrate in a physical vapor deposition (PVD) chamber includes applying RF power at a VHF frequency to a target comprising a metal disposed in the PVD chamber above the substrate to form a plasma from a plasma-forming gas; optionally applying a DC power to the target to direct the plasma towards the target; sputtering metal atoms from the target using the plasma while maintaining a first pressure in the PVD chamber sufficient to ionize a predominant portion of the sputtered metal atoms; and controlling the plasma sheath voltage between the plasma and the substrate to form a metal-containing layer having a desired crystal structure and or desired morphology on feature structures.

It is an object of the present invention to provide a method and apparatus for locating a keyword of a speech and a speech recognition system. The method includes the steps of: by extracting feature parameters from frames constituting the recognition target speech, forming a feature parameter vector sequence that represents the recognition target speech; by normalizing of the feature parameter vector sequence with use of a codebook containing a plurality of codebook vectors, obtaining a feature trace of the recognition target speech in a vector space; and specifying the position of a keyword by matching prestored keyword template traces with the feature trace. According to the present invention, a keyword template trace and a feature space trace of a recognition target speech are drawn in accordance with an identical codebook. This causes resampling to be unnecessary in performing linear movement matching of speech wave frames having similar phonological feature structures. This makes it possible to improve the speed of location and recognition while ensuring the precision of recognition.