21results about How to "Preserve precision" patented technology

The invention relates to a robot action generation method and system combining a universal model and a special model, and the method comprises the steps: constructing the universal model and the special model, carrying out the pre-training of the universal model, carrying out the parameter fine tuning of the pre-trained universal model, and training the special model based on the fine-tuned universal model; obtaining a task instruction and real-time visual information, inputting the task instruction and the real-time visual information into the fine-tuned universal model, and outputting an action sequence and task potential features; and acquiring real-time point cloud sensing data, inputting the real-time point cloud sensing data, the action sequence and the task potential features into the trained special model, and outputting continuous actions of the robot. Compared with the prior art, the robot action generation speed is increased, and the generalization of robot action generation is enhanced.

The invention provides an external wall quality defect identification system, which comprises a preprocessing module used for preprocessing an image of a to-be-detected external wall, the image comprising an infrared image and a high-definition image of the external wall; the defect judgment module is used for analyzing the image by using a pre-trained machine learning model so as to identify and mark potential defects in the image; the defect confirmation module is used for finely screening the potential defects so as to determine real defects from the potential defects; and the output module is used for outputting the image of the outer wall marked with the defect. The invention further provides a corresponding method.

The invention relates to the technical field of tire simulation design, in particular to a tire force transmissibility finite element simulation method and system and a program product. According to the method, an axial symmetry finite element model comprising a tire, a rim, internal air and an additional deformable object is established, material attributes and contact pairs are defined, and a three-dimensional tire model is generated in combination with inflation simulation. An air unit with a displacement freedom degree and a self-adaptive grid division technology are adopted. The method comprises the following steps: simulating the embedding, inflating and loading processes of a tire under different working conditions by dynamically switching a contact pair relationship, performing steady-state dynamics analysis on the force transmissibility of the tire, and calculating an acceleration frequency response function and a force transmissibility curve of a tread and a rim center. According to the method, the analysis efficiency is greatly improved, the simulation time is shortened by 80%, meanwhile, a high-precision simulation result is kept, and the method is suitable for force transmissibility evaluation of tires with multiple air pressures and multiple specifications and has wide application value.

The utility model provides a circular saw blade with positioning structures, which comprises a saw blade body provided with an assembling groove, one side of the assembling groove is provided with a first side wall, the other side of the assembling groove is provided with a second side wall, the first side wall and the second side wall are opposite to each other, the first side wall is provided with a first positioning structure, and the second side wall is provided with a second positioning structure; the blade is arranged in the assembling groove, the blade is provided with a first side face and a second side face which are opposite, the first side face is provided with a first corresponding structure, the second side face is provided with a second corresponding structure, the first corresponding structure abuts against the first positioning structure, and the second corresponding structure abuts against the second positioning structure. The saw blade mainly aims to enable the blade to be rapidly and accurately installed on the saw blade body, prolong the service life of the blade and clamp the blade without being pressed by additional screws.

This invention provides a flexible production line control method and system based on a large vertical domain model. The method includes: constructing a pruning index combining temporal stability and spatial similarity for the original large vertical domain model to evaluate the redundancy of the activation tensors in the intermediate layers, and performing structured pruning accordingly to obtain a sparse structure model; performing mixed-precision quantization on the pruned model: allocating quantization bits according to the sensitivity of each network layer weight to task loss, and calculating scaling factors and zeros based on the range of non-zero weights to represent sparse weights; calculating the output deviation between the quantized pruned model and the original model under the same input, and training a feedforward network with the former's output as input to fit and compensate for the output deviation; combining the quantized pruned model and the feedforward network to form a composite control model; after deployment, adding the output command of the quantized pruned model to the compensation vector during inference to generate control commands.

The invention provides a tool defect detection model construction method and a tool defect detection method, and relates to the technical field of image processing, and the tool defect detection model construction method comprises the steps: obtaining a training data set; for each first cutter image, performing feature extraction on each first cutter image by using a pre-constructed first unsupervised network to obtain a first feature map corresponding to the first cutter image; adjusting the channel dimension of the first cutter image according to a first feature map corresponding to the first cutter image to obtain a second cutter image corresponding to the first cutter image; and training the first supervised network according to the second cutter images corresponding to the plurality of first cutter images to obtain a second supervised network, and taking the second supervised network as a cutter defect detection model. Compared with the prior art, the accuracy of tool defect detection can be enhanced.

This invention provides a fast calculation method for beam transport in a de-energizer section based on the transfer matrix and Monte Carlo method, relating to the field of beam physics. The method includes: Step 1, constructing a transfer matrix model of the beam transport section before the de-energizer based on the beam optical parameters and optical element parameters at the starting point of the beam transport line, obtaining the particle phase space distribution parameters at the de-energizer inlet; Step 2, establishing a three-dimensional structural model of the de-energizer based on the particle phase space distribution parameters at the de-energizer inlet and performing Monte Carlo beam transport simulation of the de-energizer section, obtaining the position, momentum, and energy information of the beam particles at the de-energizer outlet; Step 3, performing outlier data removal processing on the position, momentum, and energy information of the beam particles at the de-energizer outlet, obtaining processed particle data. This invention achieves high-precision reconstruction of Monte Carlo particle information into optical parameters, improving the simulation efficiency of beam transport lines containing de-energizer structures.

The invention belongs to the field of unmanned aerial vehicle aerodynamics, and particularly relates to a blade fluid-solid coupling method based on a DMD reduced-order model, and the method comprises the following steps: obtaining key parameter variables of blade deformation, and setting a variation range corresponding to each key parameter variable according to constraint conditions formed by the configuration of the key parameter variables; establishing a sampling space based on key parameter variables by adopting a Latin super-cubic method; through fluid simulation software, a CFD simulation method is adopted to obtain flow field aerodynamic force data of a flow field on the sampling space; establishing a flow field price reduction model in a sampling space by adopting a DMD dynamic mode decomposition method through simulation software according to the flow field aerodynamic data; according to the flow field price reduction model, a velocity field and a pressure field of the flow field are obtained, the initial aerodynamic force of the paddle is obtained through integral calculation, and the initial aerodynamic force of the paddle is guided into the CSD model; and on the basis of the flow field reduced-order model, considering the condition of elastic deformation of the blade, solving the aerodynamic force of the blade again in a loose coupling manner in combination with the blade CSD model until convergence, and extracting the aerodynamic force of the blade.

The present invention provides a method, device and computer equipment for constructing a three-dimensional fracture network rock model. The method includes: obtaining geometric information of multiple planes of the rock mass; obtaining intersection information of intersection segments formed by each plane intersecting with all other planes; adjusting each plane and each intersection line to a new position by rotation and translation; generating a first grid surface based on fractal theory, and using the first grid surface to perform grid fluctuation attenuation processing on multiple intersection segments of each plane to obtain multiple second grid surfaces; cutting each contour grid surface with the contour intersection line as the cutting line to obtain multiple contour fractal surfaces; adjusting multiple contour fractal surfaces and multiple joint grid surfaces back to their original positions by rotation and translation to obtain a rock model to be cut; cutting the rock model to be cut based on the internal intersection line to obtain multiple block models. The present invention uses the fractal theory principle to dynamically adjust the grid characteristics of the intersection area, solving the problem of difficult grid division of multiple fractal surfaces.

The embodiment of the invention provides a data cutting method and device, electronic equipment and a storage medium, and the method comprises the steps: selecting a target sample from a plurality of samples of a sample data set, and obtaining a target label of the target sample; determining a plurality of neighborhood samples of the target sample in the sample data set, and obtaining an adjacent feature matrix according to feature codes of the plurality of neighborhood samples; performing feature propagation processing on the adjacent feature matrix based on the feature codes to obtain an updated adjacent feature matrix, and calculating an adjacent predicted value of each neighborhood sample based on the updated adjacent feature matrix; based on the adjacency prediction value and the label probability value of the target label, calculating to obtain a similarity estimation value between the plurality of neighborhood samples and the target sample; and carrying out data cutting on the sample data set based on the similarity estimation value to obtain a core data set, and carrying out label re-labeling on the target sample of the core data set based on the similarity estimation value so as to improve the accuracy of carrying out data cutting in the sample data set to retain important sample data.

The application relates to an intelligent content processing system based on a large model and a semantic vector, which comprises a user content preprocessing module, a user content search service module, a user content access processing module, a user content recommendation module, a user comment, opinion and dynamic classification and statistics module of a scroll bar and a user sensitive word and sentence analysis module and the like. The application also relates to a processing method. The intelligent content processing system based on the large model and the semantic vector and the method thereof adopt the application, automatically extract the abstract, the main body and the theme of the content through the large model, generate the semantic vector, combine the traditional inverted index to build a hybrid retrieval service and realize the deep fusion of keyword matching and semantic understanding. Meanwhile, the interest model is dynamically constructed based on the user behavior, various strategies are fused to realize accurate personalized recommendation, a double-layer mechanism of DFA rule and semantic vector matching is adopted for intelligent content auditing and classification, and the intelligent level, the accuracy and the system efficiency of the content processing are significantly improved.

The invention relates to a medical beauty AI GC intelligent large model system and a control method thereof, the system comprises a data acquisition module, a preprocessing and differential privacy module, an AI GC generation module, a data security and protection module and a control module, the data acquisition module acquires medical data and beauty demand information of a patient through a sensor and a camera; the preprocessing and differential privacy module is used for cleaning and standardizing original data, adding noise to protect user privacy by applying a differential privacy technology, and dynamically adjusting the noise amount by adopting an algorithm. The system can dynamically adjust the noise amplitude according to the sensitivity of the data, so that the efficiency and the individuation level of the medical beauty service are remarkably improved, powerful safety guarantee is provided for the medical beauty data, and the problems of privacy disclosure and safety of the AI technology in medical data processing are effectively solved.

The present disclosure provides a video encoding and decoding method, device, equipment and medium for machine vision tasks, and relates to the field of machine vision technology. The method includes: obtaining visual task information corresponding to the video image to be encoded; judging whether the video image to be encoded meets the image preprocessing conditions before encoding according to the visual task information corresponding to the video image to be encoded; if the video image to be encoded meets the image preprocessing conditions before encoding, performing a preprocessing operation on the video image to be encoded, and encoding the video image after the preprocessing operation; if the video image to be encoded does not meet the image preprocessing conditions before encoding, directly encoding the video image to be encoded. The present disclosure reduces the redundancy caused by the difference in image size between the source and the input layer of the downstream visual task network, and retains the accuracy of subsequent machine vision tasks while reducing the redundancy.

The invention relates to the technical field of computer visual privacy protection, and discloses a visual privacy protection method and device for video data, electronic equipment and a storage medium, and the method comprises the steps: compressing to-be-processed video data, and determining compressed video data comprising a plurality of I frames, a plurality of P frames and a plurality of B frames; for any I frame, dividing the I frame into a plurality of areas according to a preset window scale; according to a gray level histogram corresponding to any region in the I frame, determining a privacy measurement entropy corresponding to the region; encrypting the I frame according to the privacy measurement entropy corresponding to each region, and determining an encrypted frame corresponding to the I frame; and determining target video data according to the encrypted frame corresponding to each I frame, the plurality of P frames and the plurality of B frames. According to the embodiment of the invention, the visual information complexity of the video frame can be quantitatively evaluated, the encryption processing is performed in a targeted manner, and the availability of the video data is ensured while the visual privacy protection effect is improved.

The application provides a kind of partition flood control early warning method and system of fusion multi-source data, it is related to flood control early warning technical field, the method includes: obtaining the multi-source monitoring data set in target river basin;The multi-source monitoring data set is preprocessed, and the standardized data set is obtained;Based on the standardized data set, the flood evolution characteristic sub-model of each partition in target river basin is respectively constructed;Lightweight physical-empirical hybrid model is used to the output result of the flood evolution characteristic sub-model of each partition is coupled calculation, and the coupling calculation result is obtained;Based on machine learning algorithm analysis historical flood case data, generate the dynamic early warning threshold that is adapted to the flood evolution characteristic sub-model of each partition;Based on the coupling calculation result and the dynamic early warning threshold, generate partition early warning information.The application can fuse multi-source data, be aimed at partition characteristic, dynamically adjust threshold and efficiently calculate.

The present invention discloses a method for weighted average fast forward back-generation robust state estimation of a radiation network. The present invention replaces the traditional least square state estimation method by adopting a weighted averaging method, and adopts a forward back-generation idea from local to global. First, node power estimation is performed on the end nodes of the branch, and then a weighted average state estimation is performed on the branch. After the estimation of all nodes and branches is completed from the bottom branch to the top branch, a correction value of the estimated value of the system state quantity is obtained from the top to the bottom, thereby realizing the weighted average fast forward back-generation robust state estimation of the radiation network.

The present application relates to the technical field of industrial numerical control, and more particularly to a numerical control system adaptive vibration suppression method, system, device and medium, the method constructs a static stiffness attenuation factor representing the spatial posture of the machine tool, establishes a process damping speed threshold model in combination with the instantaneous cutting depth; the benchmark frequency is modified by introducing the thermal drift and the force-induced softening amount for multi-physical field coupling, and the notch depth is dynamically solved by using a nonlinear gain function according to the speed deviation, and the filter is reconstructed synchronously; the technical problems that the existing technology cannot identify the physical stability boundary, leading to false triggering of the filter, and ignoring the thermal-force coupling, leading to frequency off-target, are solved; the low-speed process damping characteristics are used to actively bypass the filter in the self-stable zone, the dynamic bandwidth of the servo system is maximized, and the frequency deviation caused by temperature rise and load can be compensated in real time, ensuring that the vibration suppression strategy accurately locks the resonance point, and significantly improving the machining surface quality and production efficiency under complex working conditions.

This invention discloses a method for joint synchronization of long-sequence phase of wireless signals and its ISAC action recognition method and device, belonging to the interdisciplinary field of radar signal processing and wireless communication. This invention aims to solve the problem of high-performance sensing with low-cost hardware in ISAC. Specifically, multi-carrier frequency-shift linear frequency modulation technology is used at the transmitting end to embed communication data while maintaining the characteristics of the radar detection waveform. To address the random carrier frequency offset, random sampling frequency offset, and random sampling time offset introduced by hardware non-ideals, a joint synchronization neural network is designed. A long-sequence feature extraction module captures long-sequence phase drift, achieving high-precision parameter decoupling and compensation in a single operation. Subsequently, the synchronized single-channel time-frequency features are combined with an environment-aware feature fusion network, and an adaptive gating fusion mechanism is used to eliminate static environmental clutter, achieving highly robust action gesture recognition.

The application discloses a nonlinear function quantization method in a Transformer compiler. The application adopts a quantization method to quantize and compress a transformer and a derivative model thereof, and obtains quantized weights and quantized parameters through a quantization-aware training (QAT) method. In the quantization process, a segmented linear approximation method is used to calculate a nonlinear function. In the determination of the segmented boundary, the QAT method is also used, and the linear approximation function of the nonlinear function is expressed as a differentiable function, so that it participates in the back propagation of the QAT, thereby obtaining the optimal segmented boundary combination. The application can be applied to the multi-platform deployment of a BERT model, and is especially suitable for a deployment platform such as FPGA which is suitable for rule calculation. In the quantization process, the method of the application can improve the hardware efficiency and maximize the accuracy.

The present invention provides an adaptive coordinated control loading and unloading method based on rock fracture information, which relates to the technical field of rock mechanics testing. The acoustic emission sensor of the present invention monitors the rock fracture signal and converts it into an electrical signal, and transmits the fracture signal to the acoustic emission acquisition device. The acoustic emission acquisition device processes and analyzes the acoustic emission signal and transmits it to the PLC controller. The PLC controller calculates and outputs the deformation control signal through the control algorithm, thereby realizing adaptive adjustment of the input control command. After receiving the deformation control instruction, the EDC controller adjusts the force acting on the rock sample through the feedback control test machine servo actuator to realize stable control of the brittle fracture process of the rock sample.

The invention relates to a block contact search method based on multi-coverage proximity search and local block entry, which aims at the problem of low calculation efficiency of a contact search method based on a contact theory, and comprises the following steps of: screening adjacent blocks and adjacent geometric elements by utilizing a multi-coverage proximity search method in a coarse search stage; in the fine search stage, a local entry block is only generated between adjacent geometric elements by using a contact theory; according to the new concept of locally entering the block, only part of geometric information of the block is utilized to participate in contact search, so that the contact search time is greatly shortened while the calculation precision is ensured. Compared with the prior art, the method can provide accurate and rapid contact search results for discontinuous numerical simulation of various complex rock mass projects such as convex bodies and concave bodies.