51results about How to "Reduce computing power requirements" patented technology

The invention discloses a permanent magnet synchronous motor sensorless control method based on a dynamic position error. A transition stage is provided, per-unit estimation position errors between a first position error signal obtained based on a high-frequency signal injection method and a second position error signal obtained based on a sliding mode observer are compared, an optimal switching speed point is dynamically determined, smooth fusion is carried out on the two position errors by using a nonlinear weighting function, and the optimal switching speed point is obtained. And finally, the estimated position and rotating speed are obtained through a phase-locked loop. And finally, position tracking is carried out by adopting different error signals according to the estimated rotating speed. According to the method, the problem of estimation value pulsation and jump caused by the fact that a traditional method depends on fixed experience switching points is solved, smooth and self-adaptive switching of the two estimation methods in the full-speed domain range is achieved, the control precision and operation stability of the system are remarkably improved, the structure is simple, and engineering implementation is easy.

The invention belongs to the crossing field of artificial intelligence and barrier-free assistance technology, and particularly discloses a sign language translation method and device based on a large language model.The method is applied to a terminal and comprises the steps that key points are extracted from sign language video frames; based on the coordinates, types and global indexes of the key points in the sign language video frames, obtaining fusion features through feature fusion; based on the fusion features corresponding to the sign language video frames in the frame window, time sequence modeling and classification are carried out, and a word prediction result of the frame window is obtained; a sliding window voting and deduplication mode is adopted, the candidate word sequence is preprocessed, a preprocessed word sequence is obtained, and the candidate word sequence is formed by connecting word prediction results of all frame windows in series according to a time sequence; based on the preprocessed word sequence, word error correction and statement generation are carried out by calling a large language model, and a sign language translation result is generated. According to the invention, sign language video translation can be realized for the terminal, and the accuracy and fluency of the translation result can be guaranteed.

This invention discloses a capture device for unmanned underwater vehicles (UUVs), comprising a submersible fixed by a mounting frame. The mounting frame is symmetrically mounted with a grasping mechanism for capturing the UUV via a base. The grasping mechanism includes a drive arm driven by the base, with the other end of the drive arm rotatably connected to four corner support frames. The four corner support frames are driven by connecting arms to contact claws for gripping. A groove is provided on the side of the drive arm near the base, and a servo motor is mounted on the base. The servo motor is connected to a gear at one end of a limiting link, transmitting power through the other end of the limiting link to the groove on the drive arm to achieve the movement of the grasping mechanism. This invention also discloses a method for launching and recovering UUVs in conjunction with the above-mentioned launch and recovery device. This invention simplifies precise hook docking operations in high sea states, reduces the dependence of UUVs on computing power during launch and recovery, and improves the automation level of the launch and recovery process.

This application discloses a method for determining the radiated temperature received by a human body, an air conditioner, and a control method thereof. The method for determining the radiated temperature received by a human body includes: determining the current pose information of a human body indoors; determining the visual factor between the current pose and the indoor surface based on the correlation between the current pose information, preset pose information, and the visual factor between the preset pose information and the indoor surface; and determining the radiated temperature received by the human body based on the visual factor between the current pose and the indoor surface and the indoor surface temperature. The air conditioner can be an energy-saving air conditioning device. This application has the advantages of real-time operation, low cost, low computing power requirements, and energy saving.

The application relates to a road surface damage detection method based on local gray extremum points and feature fusion, which comprises the following steps: dividing a road surface image into a plurality of windows, and respectively calculating local extremum points in the window units; determining a plurality of local extremum points as feature points according to the gray distribution of the neighborhood blocks of the extremum points; intercepting the image blocks of the neighborhoods of the local extremum points as the centers, establishing a block image data set, and training a CNN block image feature extraction model; using the trained CNN feature model to calculate the feature vectors corresponding to the feature points of the image windows, performing feature fusion to obtain a composite feature vector; and using a support vector machine to classify the composite feature vector to detect the damage of the road surface. The application has the beneficial effects that the application combines some prior image features of the damage target area with a deep learning algorithm, reduces the scale of the data set and the requirement of the calculation capacity, and improves the detection effect of the road surface.

The application relates to a kind of star-ground cooperative resource scheduling methods, comprising: satellite side receives terminal uplink data packet and analysis, adopts dynamic sliding window to extract physical layer feature and service layer key field, runs quantized compression isolated forest model, generates exception event label and is associated with original data packet after, as observation data is downloaded to ground side;Ground side receives observation data, combines historical data to form multi-star joint observation data, constructs space-time cubic grid, based on space-time cubic grid, carries out space-time correlation analysis to multi-star joint observation data, calculates the space-time correlation coefficient of abnormal event time sequence and spatial aggregation degree sequence, when space-time correlation coefficient matches preset threshold value, identify event type and generate structured alarm;According to event type, generate adaptive scheduling strategy, upload scheduling strategy instruction to satellite through ground side and satellite's instruction upload link, and resource allocation is executed according to scheduling strategy by satellite when overtop.

The invention discloses an unmanned ship autonomous cluster formation obstacle avoidance navigation method. The method mainly comprises the following steps: initializing parameters, position, course and operation plan of each ship in an unmanned ship cluster; in the sailing process, the pilot boat and the following boat share sailing information to the cluster network; in the sailing process, sea area environment information is detected according to a sensing module of the unmanned ship, and whether an obstacle is encountered or not is judged; when an obstacle is encountered, the unmanned ship formation carries out control behavior conversion, and corresponding obstacle avoidance measures are taken. According to the method, through obstacle avoidance path planning of a pilot boat, local collision avoidance of following boats based on a dynamic window, and ranking according to formation parameters during obstacle avoidance of the whole formation, when an unmanned boat cluster encounters an obstacle, respective optimal control behaviors can be flexibly adopted for collision avoidance on the basis of guaranteeing safety and accessibility, and the unmanned boat cluster is more accurate in obstacle avoidance. And moreover, rapid aggregation can be realized after obstacle avoidance, cluster navigation can be continuously kept, and the obstacle avoidance efficiency is effectively improved.

This invention discloses a humanoid robot mechanical finger, comprising a fixed frame with a first directional tooth; a proximal phalanx drive rod with one end movably connected to the output end of a telescopic device and the other end having a second directional tooth, and a distal phalanx linkage rod rotatably connected thereto; a middle phalanx linkage rod rotatably connected to a second position of the fixed frame; a driven gear assembly rotatably connected between the proximal phalanx drive rod and the middle phalanx linkage rod, and meshing with the first directional tooth; a middle phalanx drive rod rotatably connected between the proximal phalanx drive rod and the middle phalanx linkage rod, and having a third directional tooth, which meshes with the driven gear assembly; and a distal phalanx swing rod rotatably connected between the distal phalanx linkage rod and the middle phalanx drive rod, and having a fourth directional tooth, which meshes with the second directional tooth. This invention employs pure mechanical gear transmission, without relying on elastic elements, wire ropes, rubber seals, or other components that are prone to aging or susceptible to temperature / radiation effects, and can operate stably for a long time in extreme environments such as high temperature, low temperature, vacuum, and radiation.

The embodiment of the invention discloses a plane X-ray imaging detection method and system. The method comprises: irradiating a plurality of regions of a human body with planar X-rays; acquiring energy projection data of a plurality of energy windows of a plurality of areas of the human body by using a photon counting detector; obtaining image information of a corresponding human body area according to the energy projection data of the plurality of energy windows; and splicing the multiple pieces of image information to obtain a detection image. According to the planar X-ray imaging detection method and system disclosed by the embodiment of the invention, abundant detection information can be conveniently and accurately obtained under the combined action of the planar X-ray and the photon counting detector.

The present application relates to the technical field of robots, and in particular to a motion generation method, system, device and storage medium for a humanoid robot, comprising: physical simulation rendering generates a visual video containing visual representation of joint motion state of the robot; analyzing the visual video to obtain frame images, natural language instructions and motion sequences, decoding output trajectory data; and outputting control instructions through a hierarchical architecture of high-level motion planning and bottom-level dynamics tracking to drive the robot to perform actions. The present application implicitly encodes the joint state of the humanoid robot in the visual features, improves the inference efficiency and generalization ability of the model, and the hierarchical decoupling architecture of the large model high-level trajectory planning and the reinforcement learning bottom-level dynamics tracking controller pays a small performance cost, which can compress the inference time delay to within 50ms and the parameter quantity to below 2 billion, and balances the high generalization of the stateless VLA and the small model, few-step inference, edge low latency, and low power requirement.

Embodiments of the present specification disclose an image defogging method and device, electronic equipment and storage medium. The method comprises acquiring an original format image, and performing exposure gain compensation on the original format image to obtain a color reconstruction image. Then, the R channel component of the original image is extracted to obtain a first image corresponding to the R channel component. Further, the color reconstruction image is decomposed based on a wavelet transform algorithm to obtain a contour layer image and a detail layer image. Finally, the first image and the contour layer image are fused to obtain a fused image, and the fused image and the detail layer image are processed by inverse wavelet transform to obtain a defogging image. Through the above image processing steps, the object signal intensity in the finally obtained defogging image is high, so that even in a heavy fog shooting environment, the defogging method can achieve an ideal defogging effect, and the method has low complexity, good real-time performance, low requirements for device hardware and computing power, and meets the requirement of high applicability.

The application provides an AMT clutch wear state online diagnosis method and system, and belongs to the technical field of clutch wear diagnosis. The method comprises the following steps: triggering wear state diagnosis in the sliding film stage, the combination stage and the full combination stage of the clutch, and acquiring vibration acceleration original signals collected continuously in time at the AMT clutch pressure plate; extracting time domain statistical features related to transient impact components and energy levels in the vibration acceleration original signals, and constructing a multi-dimensional feature vector; and inputting the multi-dimensional feature vector into a pre-trained support vector machine to obtain a diagnosis result of the clutch wear state. The method analyzes the time domain statistical features of the clutch vibration signals, uses the trained vector machine model to realize online diagnosis and classification of the clutch wear state, and achieves the purpose of real-time and accurate AMT clutch fault diagnosis.

This invention relates to a federated learning training system and method based on layered privacy protection. Distributed clients deployed locally by multiple participants perform model training based on local data, outputting encrypted model parameters. A federated coordination center receives and aggregates the encrypted parameters uploaded by the clients, generates and distributes a global model, and the application service layer provides training services to end users based on the aggregated global model. The privacy preprocessing module of the distributed clients implements differentiated privacy protection during the training process based on data sensitivity, and the adaptive privacy budget allocation module of the federated coordination center dynamically adjusts the privacy protection strength of each client. This invention significantly reduces the risk of data leakage, meets compliance requirements, significantly improves the accuracy of personalized recommendations and skills assessment, greatly enhances the coherence and operability of the entire training process, effectively reduces hardware investment costs, breaks down resource barriers between institutions, lays the foundation for long-term cross-institutional collaboration, and continuously iterates and optimizes system performance.

The invention relates to the field of photovoltaic technology, and discloses a photovoltaic power station power prediction method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining the operation data and environment data of a photovoltaic power station; calculating a target attenuation coefficient of the photovoltaic power station according to the operation data and the environment data; and correcting an initial power predicted value of the photovoltaic power station based on the target attenuation coefficient to obtain a target power predicted value, the initial power predicted value being determined according to meteorological data of the photovoltaic power station. The method solves the problems that in the prior art, the interaction effect between photovoltaic panel aging and dust deposition is not considered, and due to the fact that a complex model or newly-added hardware is depended on, the power prediction precision of an old photovoltaic power station is low, and the applicability is poor.

This application discloses a method, apparatus, device, and storage medium for model search. The method includes: acquiring a search space, which includes a convolutional module and a visual transformation module; wherein the visual transformation module includes a self-attention sub-module for executing a self-attention algorithm with a complexity of O(n); preprocessing labeled training images using the convolutional module; performing target task-based image processing on the preprocessed results using the visual transformation module; determining a first gradient of the search space based on a first loss between the image processing results and the labels of the training images; updating the search space based on the first gradient to obtain a target search space; wherein the target search space is used to search for a target model, and the target model is used to perform target task-based image processing on the target image to be processed. This allows the searched model to have linear complexity, effectively solving the problem of difficult model deployment.

The invention aims to provide an industrial interconnection system sensor fault tolerance control method based on artificial intelligence, and the method comprises the steps: building a low-constraint switching architecture with a direction perception capability, and removing the unnecessary shutdown waiting constraint in a conventional method; a state observer and a self-adaptive compensation mechanism are utilized to reconstruct a real signal through a software algorithm under a sensor fault; predetermined performance control is realized through an error preprocessing mechanism; an unknown item of the system is approached by using an artificial intelligence technology based on a fuzzy logic system; and a dynamic surface signal processing technology is introduced, so that the computing power demand of the controller is greatly reduced. Therefore, efficient, safe and low-cost operation of the industrial interconnection system under complex working conditions is realized.

This invention discloses a method, system, device, and storage medium for virtual-real fusion based on spatial positioning. The method includes: determining multiple spatial positioning base points in a real-world scene and determining the relative positional relationship between each spatial positioning base point and the physical object to be interacted with; constructing a first coordinate system in a virtual digital space and determining the first coordinate of each spatial positioning base point in the first coordinate system; modeling the real-world scene according to the first coordinate system to obtain a virtual digital model, and then determining the model area to be fused in the virtual digital model according to the relative positional relationship and the first coordinate; and performing virtual-real fusion between the model area to be fused and the physical object to be interacted with based on the spatial positioning base points. This invention reduces the number of comparisons and identifications between physical objects and digital models, improving the efficiency of mixed reality display and interaction; it also ensures the real-time and preparedness of virtual-real fusion, greatly enhancing the user's interactive experience, and can be widely applied in the field of mixed reality technology.

The invention discloses a video compression method based on motion estimation and affine transformation substitution. The method comprises the following steps of: 1, performing feature analysis and initial coding on an image to realize initial compression of an original video frame, and establishing spatial geometric association between adjacent frames; 2, mapping high-resolution homography to a low-resolution coding space by using a down-sampling matrix based on spatial geometric correlation to obtain a down-sampling homography matrix; step 3, according to the mask region of the moment coding data obtained by calculation in the step 2, separating to obtain a code of a cutting background, namely a code of an increment background, and realizing differential compression; and 4, according to the codes of the incremental background obtained in the step 3, the decoding end combines the stored codes at the T-1 moment to splice and fuse reconstructed coding data at the T moment, and then video frames are decoded. According to the invention, the adaptive compression coding capability for different regions in the video compression process is effectively improved.

Embodiments of the present application disclose a road edge detection method, an electronic device and a storage medium. The electronic device obtains road point cloud data; performs edge key point extraction processing based on the road point cloud data to obtain edge key point information corresponding to the road point cloud data; determines at least one point cluster based on the edge key point information, and determines at least one straight line segment information according to the at least one point cluster; wherein the at least one straight line segment information is determined according to a target unit vector corresponding to each of the at least one point cluster; and performs aggregation processing according to the at least one straight line segment information to obtain a road edge detection result.

This application provides a method, apparatus, device, and storage medium for identifying a target object, relating to the field of video processing technology. The method includes: when identifying a target object, first acquiring video frame images of the target object captured by a vision device, and simultaneously detecting the position of the target object by a radar device; and based on the position of the target object, determining the target image region where the target object is located from the video frame images; based on the target image region, extracting a portion of the video frame images containing the target object; and then identifying the target object in the portion of the video frame images to obtain the identification result of the target object. Using the technical solution of this application, the accuracy of the identification result can be improved without being limited by computing power.

This invention relates to the field of digital image processing technology, and discloses a digital image processing method and system for video. The method involves: acquiring multiple image data streams and performing coordinate transformation and spherical projection to obtain a panoramic stitched image; extracting a preset set of bounding boxes from the panoramic stitched image and performing target recognition to obtain a first detection position of the ball target and a second detection position of the player target; calculating the first real-time position and first velocity of the ball target, and calculating the second real-time position and second velocity of the player target; determining the center coordinates and screen range of the broadcast view, and generating an automatically directed video stream based on the center coordinates and screen range of the broadcast view. This invention effectively solves the stitching seam problem caused by depth parallax in traditional panoramic stitching algorithms based on feature point matching by fixing and calibrating feature points on the court surface and utilizing the single-plane characteristic of the court surface.

The invention belongs to the technical field of aircraft target allocation and planning, and particularly relates to a heterogeneous fixed wing cluster target allocation and planning method oriented to flight time collaboration, which comprises the following steps: firstly, introducing a KMEANS method, and clustering targets based on target positions to form target clusters corresponding to unmanned aerial vehicles; then, initial planning of the route in each target cluster is realized based on an improved genetic algorithm; and finally, redistributing and replanning targets in each target cluster on the basis of target position characteristics aiming at the endurance constraint caused by mutual difference of the optimal cruise speeds, so as to realize further reduction of endurance and endurance collaboration. The time-of-flight cooperative target allocation and planning algorithm is low in time complexity, low in computing power requirement and easy to deploy, and can be widely applied to top task planning and target decision allocation tasks of various heterogeneous unmanned aerial vehicle clusters in the future.

The invention provides a fusion reactor CICC superconducting magnet equivalent winding material parameter calculation method, and the method comprises the steps: determining a minimum unit cell unit corresponding to a superconducting magnet winding; constructing a finite element model of a minimum unit cell unit; applying a specific load and a constraint condition to the finite element model, outputting a calculation parameter corresponding to the initial material parameter, and calculating the initial material parameter by using a generalized Hooke's law according to the calculation parameter; establishing an equivalent model of the superconducting magnet winding according to the initial material parameters, and performing comparison verification on the equivalent model and a pre-established detailed model of the superconducting magnet winding; wherein if the comparison verification is passed, the initial material parameter is used as the material parameter of the equivalent winding corresponding to the superconducting magnet winding; and if the comparison verification is not passed, re-determining a new minimum unit cell unit. According to the invention, the minimum unit cell units replace the whole winding, and the winding is simplified into the set of the minimum unit cell units, so that the calculation efficiency can be improved while the equivalent parameter precision is ensured.

This invention provides a method for measuring the volume of a cuboid based on a single image, relating to the field of dimensional measurement technology. The method includes fixing a monocular camera and capturing an image of a calibration plate, obtaining the translation vector of the camera relative to the calibration plate in that pose, calculating the height of the camera's coordinate system origin above the ground, capturing an image containing one lower vertex of the cuboid and its corresponding length, width, and height, and picking the pixel coordinates of the four vertices of the cuboid. Combining the intrinsic parameters of the monocular camera, the height above the ground, and the pixel coordinates of the four vertices of the cuboid, the length, width, and height of the cuboid are calculated using geometric knowledge, thereby obtaining the volume of the cuboid. This invention requires only a single image, is low-cost, requires minimal computing power, and is easy to implement.

The embodiment of the invention provides a logistics tracking method and product, electronic equipment and a storage medium, and belongs to the technical field of block chains. The method comprises the following steps: acquiring an uplink request of data acquisition equipment, adding the data acquisition equipment into a block chain system under the condition that the uplink request passes verification, and constructing the data acquisition equipment as a light node; constructing a logistics management system as a consensus node, and controlling the logistics management system to perform uplink processing on the logistics data sent by the data acquisition equipment; wherein the logistics management system is a consensus node; sending an uplink result generated by the logistics management system to the data acquisition equipment; and under the condition that the data acquisition equipment receives the preset number of uplink results, controlling the data acquisition equipment to exit the block chain system. According to the embodiment of the invention, the data uplink risk can be reduced.

The application discloses a pose estimation method, related equipment and a storage medium, wherein the method comprises the following steps: obtaining a relative pose quantity of a to-be-measured object at a tth acquisition time based on inertial measurement unit (IMU) data of the to-be-measured object at the tth acquisition time and at a previous time; obtaining a pose prior quantity at the tth acquisition time based on the relative pose quantity and a pose optimization quantity at a pth acquisition time; wherein t and p are positive numbers, and the pth acquisition time is a previous acquisition time of the tth acquisition time; obtaining a pose estimation quantity of the to-be-measured object at the tth acquisition time based on the pose prior quantity at the tth acquisition time and an observation quantity at the tth acquisition time; and positioning the to-be-measured object based on the pose estimation quantity of the to-be-measured object at the tth acquisition time. Lightweight pose estimation can be realized.

This application proposes a speech decoding processing method, apparatus, electronic device, and storage medium. When a blank decoding candidate exists among all decoding candidates corresponding to the current audio acoustic output of an audio recognition system, the method determines whether the current audio acoustic output is a blank output. If it cannot be determined that the current audio acoustic output is blank, this application does not employ a language model where each node has a self-jumping arc for self-jumping. Instead, it creates a new activation arc for self-jumping on the active node with blank decoding candidates in a pre-defined language decoding network, and adds blank decoding candidate information as a label to the activation arc. Based on the activation arc label, the path label of the active node is determined, thereby reducing the memory usage of the self-jumping arc portion. Based on this, this application reduces the memory and computing power requirements of speech recognition, mitigating the limitations of speech recognition technology in practical applications.

The invention provides a charging control method and charging equipment. The charging control method provided by the invention comprises the following steps: when request charging information is received, sending a charging strategy request to cloud equipment according to the request charging information; determining reference charging power of the charging equipment based on a charging strategy fed back by the cloud equipment; acquiring actual grid-connected power between the alternating-current bus and the power grid and actual charging power of the charging equipment through the intelligent electric meter; determining redundant power according to the actual grid-connected power and the actual charging power; determining target charging power of the charging equipment according to the reference charging power and the redundant power; and generating a control signal according to the target charging power and sending the control signal to the charging equipment, wherein the control signal is used for adjusting the actual charging power of the charging equipment. According to the charging control method provided by the invention, the charging cost can be better reduced while the charging requirement of a user is met.

This invention discloses a method and device for rapid synchronous phasor sensing of single-phase current in power system transmission and distribution lines, belonging to the field of current measurement technology. The method includes: constructing a current sensing probe using TMR chips and measuring the spacing between the chips; placing a printed circuit board carrying the current sensing probe at the line under test and acquiring the instantaneous output voltage values ​​of each TMR chip; calculating the phase of the current phasor under test based on the instantaneous output voltage values ​​of the TMR chips; and calculating the amplitude of the current phasor under test based on the instantaneous output voltage values ​​of the TMR chips and the spacing between the chips. The method and device for rapid synchronous phasor sensing of single-phase current in power system transmission and distribution lines described in this invention can solve the problems of poor installation convenience and complex calculation processes in existing current phasor sensing methods.