110 results about "Fast algorithm" patented technology

The invention discloses an emotion recognition method based on electroencephalogram (EEG) and expression data, which combines fusion analysis of electroencephalogram (EEG) signals and facial expression data, and provides an innovative solution for the complexity and accuracy of emotion recognition. The method comprises the following specific implementation steps: 1, finding an optimal frame in an expression video by using multi-instance learning, extracting global visual features by adopting Swin Transform, and realizing dimension reduction and balance representation of cross-modal information through a cross attention mechanism; 2, preprocessing the electroencephalogram signal by using band-pass filtering and a fast ICA algorithm, and extracting effective features by using PSD; and 3, designing a fusion module based on Transform, combining the characteristics of the two modes, and giving full play to the complementary advantage of the modes. Experimental results show that 94.85% of classification accuracy is achieved in four types of emotion classification tasks, and compared with an existing method, the accuracy is remarkably improved.

The invention belongs to the technical field of simultaneous localization and mapping in the aspect of computer vision, and particularly relates to a dynamic simultaneous localization and mapping (SLAM) method based on 3D Gaussian, which comprises the following steps of: aligning an RGB image captured by a sensor with a depth image, and processing the RGB image by using a semantic segmentation algorithm to obtain a depth image; setting a semantic tag for a pixel point where the semantic target is located; for the RGB image, an ORB feature point and a SuperPoint feature point are extracted by using a FAST algorithm and a SuperPoint algorithm respectively; feature points with semantic tags are recognized and removed, and then pose estimation of the current frame is optimized in combination with the depth image; projecting a semantic target of a historical frame into a current frame image through pose transformation, recognizing a dynamic object according to the similarity of the semantic target of the historical frame and the semantic target of the current frame and the intersection-union ratio of a semantic target mask, and respectively constructing a dynamic map and a static map; and further optimizing the static map by using a 3DGS technology to obtain a dense map finally represented by 3D Gaussian.

The invention discloses an image registration method based on improved ORB feature extraction and grid feature matching. The method comprises the following steps: inputting an image to be extracted; using an improved FAST algorithm to extract feature points of the to-be-spliced image; establishing a BRIEF descriptor, and obtaining a BRIEF descriptor with a direction; according to the feature point set of the to-be-spliced image, establishing a Delaunay triangular grid taking the feature points as vertexes; judging the similarity of triangles, and obtaining similar triangles formed by the feature points; matching the similar triangle vertexes by using a Hamming distance to obtain an optimal matching feature point pair; obtaining affine matrix parameters by using an SVD method; according to the image registration method provided by the invention, the image quality is ensured, the feature extraction efficiency and the extraction precision can be improved, wrong matching is reduced, the stability is improved, the real-time performance of video stitching is improved, the waste of memory resources is reduced, the equipment cost is reduced, the working efficiency is greatly improved, and the method is suitable for popularization and application. And a better method is provided for image splicing.

The invention discloses a self-supervised clustering method for massive single cell data. The method comprises the following steps: 1) downloading single cell RNA sequencing data from a disclosed gene expression database; 2) preprocessing the single-cell RNA-seq data, wherein the preprocessing comprises filtering, standardization and feature selection; (3) carrying out preliminary clustering on the preprocessed single cell RNA sequencing data by utilizing a rapid Louvain algorithm; 4) screening representative cell points from the preliminary clustering result by using an index of clustering representativeness; 5), after representative cell points are screened out from each preliminary cluster, a Transform model is used, and a final clustering result is obtained.According to the method, the rapid Louvain algorithm and the Transform neural network are combined, the efficiency and accuracy of clustering analysis are remarkably improved, and an efficient and accurate solution is provided for analysis of mass single cell data.

The invention provides an efficient transmission algorithm for HPLC carrier communication data, which belongs to the technical field of carrier communication, and comprises the following steps of: step 1, double-link separation processing: respectively processing frame control data and load data by adopting two separated links; 2, a frame control data processing flow: carrying out improved Turbo coding on the frame control data; step 3, a load data processing flow: firstly, carrying out adaptive scrambling on load data, and adopting Turbo coding with different parameter configuration from frame control data; step 4, OFDM symbol generation and transmission: adopting an optimized fast algorithm to reduce the operand and the processing time; and step 5, receiving end processing flow: respectively adjusting frame control and load data by adopting AGC (Automatic Gain Control) and time synchronization. The algorithm is high in anti-interference capability, remarkably reduces the bit error rate, is high in transmission efficiency, can adapt to different power line channel environments and data transmission requirements, and has good universality and stability.

The invention discloses a large-span roof structure wind-induced response fast algorithm considering high-order modal quasi-static response, and the algorithm comprises the following steps: S11, building a finite element model of a large-span roof structure, and determining the modal damping ratio of the large-span roof structure according to the design condition, so as to obtain the structural dynamic characteristics of the large-span roof structure; step S12, determining a random wind load acting on the large-span roof structure through a wind tunnel test, and calculating a wind load Fourier transform function of a truncation function corresponding to the random wind load; step S13, according to the dynamic characteristics of the structure and the wind load Fourier transform function, performing calculation by adopting a modal acceleration method, obtaining a Fourier transform function of the structure displacement response; and S14, according to the Fourier transform function of the structure displacement response, performing power spectral density estimation by adopting a periodogram method to obtain a power spectral density matrix of the structure displacement response, and performing integration on the power spectral density matrix of the structure displacement response to obtain a displacement response root-mean-square of the large-span roof structure. According to the method, quasi-static response of a high-order mode is considered, better convergence is achieved compared with a traditional method, and under the condition of the same precision, calculation time and memory occupation are relatively small.

The invention belongs to the technical field of power quality disturbance detection, and particularly relates to a power quality disturbance detection method based on improved adaptive S transformation, which comprises the following steps of: designing window width control factors of two Gaussian windows, flexibly adjusting time domain window widths and frequency spectrum characteristics of the Gaussian windows, and forming a global adaptive Gaussian window; taking upper and lower limits of time-frequency two-domain resolution as constraint conditions, taking enhanced signal energy concentration ratio as a target function of an optimization scheme, and adopting an interior point method to carry out parameter solving; performing fast Fourier transform on the signal and setting a threshold value, and defining a part higher than a threshold value point as a characteristic frequency point ki; screening calculation at a non-characteristic frequency point, only performing time-frequency transformation at a characteristic frequency point ki, and constructing a fast algorithm for improving adaptive S transformation; and obtaining a final disturbance characteristic information value. According to the method, the global adaptive Gaussian window is constructed as a kernel function for improving adaptive S transformation, the effective window length and the frequency spectrum of the window function can be automatically adjusted along with the change of the detection frequency, window parameters are prevented from being frequently switched for improving the time-frequency resolution, and various power quality disturbances are efficiently and accurately detected.

The invention discloses a multi-task-oriented edge computing resource allocation method and device, equipment and a medium. The method comprises the following steps: acquiring a user terminal set, an MEC server set, a to-be-computed task group of each terminal and a sub-channel set of each server; constructing a network flow graph which takes each sub-channel as a first type of nodes, takes each user terminal as a second type of nodes, and takes a connecting line between the first type of nodes and the second type of nodes as an edge; constructing a cost calculation formula corresponding to the edge in the network flow graph; dynamically searching a task execution strategy with the minimum total cost by adopting a shortest path fast algorithm based on the network flow graph; according to the technical scheme provided by the embodiment of the invention, through joint optimization calculation unloading and resource allocation, the throughput is remarkably improved on the premise of ensuring the task time delay demand; a dynamic cost adjusting mechanism can adapt to network environment changes in real time, and efficient utilization of resources is achieved.

The invention relates to the technical field of automobile electronic intelligent fault diagnosis, and discloses a multi-modal data fusion automobile air conditioner controller abnormal state monitoring method, which comprises the following steps: deploying five types of intelligent sensor nodes at key positions of an automobile air conditioner system to realize hardware-level time synchronization and adaptive sampling; a fast ICA algorithm is combined with sliding window processing to realize signal separation, a self-adaptive wavelet basis selection mechanism is designed to carry out time-frequency analysis, and principal component analysis is used to carry out feature dimension reduction; constructing a lightweight attention fusion network, dynamically adjusting different modal feature weights, and mining feature relevance through a cross-modal interaction module; a hierarchical progressive anomaly detection architecture is designed, a statistical method and a neural network detection result are fused, a time consistency constraint mechanism is established, and multi-time scale anomaly prediction is realized; and fault type identification and severity evaluation are carried out in combination with an expert knowledge base, and a multi-priority protection strategy and an emergency plan are formulated.

The invention discloses a target detection pre-labeling method based on a fast algorithm, and relates to the technical field of computer vision and geometric processing, and the target detection pre-labeling method based on the fast algorithm comprises the following steps: S1, data acquisition and preprocessing; and the collected image data is transmitted to a cloud data center in real time through wireless transmission. According to the target detection pre-labeling method based on the fast algorithm, image information can be locked and recognized more accurately through operation of two times of pre-labeling, so that the collected image information can be recognized and screened more accurately, a certain workload can be reduced for subsequent manual auditing and correction, and the accuracy of target detection and pre-labeling is improved. And meanwhile, it is also ensured that the pre-labeling model can carry out accurate labeling processing on a relatively fuzzy or shielded image, and through operation of a YOLOv5 algorithm, the pre-labeling model can be more efficient and faster when carrying out image recognition.

The invention relates to a multi-user communication enhancement method based on a multi-metasurface network. The method comprises the following steps: deploying the multi-metasurface network; performing three-dimensional beam scanning on the plurality of programmable metasurfaces to obtain an angle amplitude spectrum; position information of each user receiving end is obtained through estimation according to the angle amplitude spectrum; constructing a directed graph; constructing a gain weight matrix according to the distance between the nodes; on the basis of the gain weight matrix, a shortest path fast algorithm is adopted, and a programmable metasurface needing to participate when service is provided for each user receiving end is determined; and for each programmable metasurface, calculating the phase configuration of beamforming of each user receiving end providing service by the programmable metasurface, and synthesizing all the phase configurations to obtain the final phase configuration of the programmable metasurface. According to the method, the problem of low coverage efficiency of a single RIS system in an indoor short-distance scene due to a far-field boundary condition is solved, the signal coverage range and the communication quality are remarkably improved, and the method is particularly suitable for a high-density Internet of Things equipment scene.

The present invention discloses an inversion method for arbitrary variable density interfaces, and specifically relates to the field of gravity exploration technology. Based on the method of three-dimensional prism partitioning, the present invention constructs a three-dimensional density interface model in a three-dimensional space grid, and by performing Taylor series expansion on the density interface forward function and retaining the first-order term, the linear function of the density interface model forward is approximately obtained and brought into the regularization objective function, and the density interface inversion iterative equation is obtained by minimizing the objective function, and the density interface inversion model is solved using a linear conjugate gradient algorithm. By iteratively correcting the density interface multiple times, the density interface and density model of the study area are inverted. The present invention realizes the accurate acquisition of density interface distribution under complex geological structures, and based on the fast algorithm of three-dimensional gravity inversion, realizes batch rapid inversion of gravity-density interfaces, laying a foundation for the exploration and development of complex geological structures.

The invention provides a low-coupling Transform target detection method and system based on FPGA (Field Programmable Gate Array) acceleration. The method comprises the following steps: constructing a Transform target detection network model based on a low-coupling optimization method; and constructing a multi-framework universal FPGA accelerator based on a Winograd fast algorithm, wherein the multi-framework universal FPGA accelerator is used for simultaneously supporting a convolutional neural network used for feature extraction and a Transform codec based on a multi-head attention mechanism in the Transform target detection network model. According to the method, an efficient memory access transformation scheme is provided for the special type of operators, so that the accelerator obtains higher convolution operation performance while having high calculation efficiency for all the operators; a high-precision approximation scheme and corresponding data flow optimization are provided for various special functions contained in an attention mechanism, a difference fitting method is provided for GELU, approximation precision is improved, attention scaling operation and Softmax operator are fused, and scaling operation overhead during operation is saved.

The present invention discloses a self-supervised clustering method for massive single-cell data, the method comprising the following steps: 1) downloading single-cell RNA sequencing data from a public gene expression database; 2) preprocessing the single-cell RNA-seq data, including filtering, standardization and feature selection; 3) using a fast Louvain algorithm to perform preliminary clustering on the preprocessed single-cell RNA sequencing data; 4) using the "cluster representativeness" indicator to screen out representative cell points from the preliminary clustering results; 5) after screening out representative cell points in each preliminary cluster, using a Transformer model to obtain the final clustering results. This method combines the fast Louvain algorithm and the Transformer neural network, significantly improving the efficiency and accuracy of clustering analysis, and providing an efficient and accurate solution for the analysis of massive single-cell data.

The invention relates to the field of new energy modeling equivalence, in particular to a wind power plant equivalence method and system based on a fast DB-SCAN algorithm and considering a low-crossing characteristic. According to the method, the time constants of the generator terminal voltage, the active power, the reactive power and the generator stator winding are selected as grouping indexes, the steady-state operation state and the transient change characteristic can be reflected at the same time, and theoretical verification is passed. Then, a traditional DB-SCAN algorithm is improved, a particle swarm algorithm is introduced for automatic optimization, the practicability of the algorithm is enhanced, a K-D decision tree is introduced for preliminary data division, the time complexity of the algorithm is reduced, and the final fast DB-SCAN algorithm has high practicability.

The invention relates to the technical field of wireless communication, in particular to an antenna radiation and mutual coupling collaborative optimization system based on a time domain hybrid fast algorithm. The antenna radiation and mutual coupling collaborative optimization system based on the time domain hybrid fast algorithm is applied to an unmanned aerial vehicle group communication system and comprises an antenna array unit, a data acquisition unit, a time domain hybrid fast algorithm unit, a mutual coupling calculation unit, an optimization control unit and a communication coordination unit. According to the invention, the three-dimensional geographic coordinates, attitude angle data and signal intensity of the unmanned aerial vehicle group are acquired in real time through the data acquisition unit, and real-time mutual coupling coefficient analysis of the mutual coupling calculation unit is combined, so that the system can dynamically sense the formation form change and trigger the optimization instruction, and the optimization control unit performs comparison through a preset threshold value. The node with the maximum deviation degree is adjusted preferentially, it is ensured that the antenna spacing and the radiation pattern are always adaptive to the dynamic environment, the mutual coupling effect is reduced, and the communication stability is remarkably improved.

The invention discloses a rail profile consistency evaluation method and system based on wheel-rail matching characteristics, and relates to the technical field of railway engineering.The rail profile consistency evaluation method comprises the steps that firstly, wheel-rail contact key indexes are analyzed, the positions of wheel-rail contact points are calculated based on three-dimensional profile data, and a model is established based on the non-Hertz contact theory to solve contact stress; a dynamic equivalent taper is obtained by using an equivalent taper fast algorithm; then, an analytic hierarchy process and an entropy weight method are combined to determine an index weight, weighted summation is performed after normalization processing is performed on original index data to obtain a comprehensive evaluation index CEI, and an evaluation standard is set in combination with a line and a train type; according to the method, multiple dynamic indexes are creatively fused to construct an evaluation system, a subjective and objective combination weighting method is adopted to overcome the defect of single weighting, the constructed quantitative comprehensive evaluation indexes provide accurate support for railway full-life-cycle management, compared with a traditional method, the evaluation accuracy is greatly improved, railway transportation safety and stability are effectively guaranteed, and the method is suitable for popularization and application. And the overall benefit and reliability are improved.

The application relates to a kind of massive SAR data imaging software design methods based on fast BP algorithm, for the existing fast BP imaging algorithm, by accurately estimating the data amount in the operation process, propose to use the available memory of operation server, dynamically correct J operator, carry out sub-scene processing to sub-set, use disk temporary storage intermediate operation result, solve the memory limit problem of SAR full scene full resolution imaging of super large data on the operation server with lower memory, greatly reduce the hardware memory cost of operation server.

The invention discloses a rapid algorithm for column design considering the most unfavorable arrangement of live loads, and belongs to the technical field of civil engineering reinforced concrete member structural design. A design control load combination in the reinforced concrete structure design is converted into a mathematical convex hull set, a universal mathematical convex hull calculation method is used for searching the design control load combination, and the most unfavorable live load arrangement combination set is obtained by calculating the convex hull set of the live load combination set; according to the convex hull merging method, two convex hull subsets calculated in batches can obtain an equivalent convex hull set through Minkowski summation and a union set of the two convex hull subsets; according to the batch calculation method, a convex hull set obtained after the convex hulls are subjected to batch enumeration calculation and convex hull set merging is completely equivalent to a convex hull set obtained after all live loads are subjected to full enumeration and then convex hull calculation.

The application provides a multi-index dynamic priority unmanned aerial vehicle cooperative path planning method, and the path planning method comprises the following steps: acquiring robot parameter information, path constraint information, task environment information and algorithm parameter information; acquiring obstacle information in a simulation environment, and constructing a simulation environment grid map, so as to uniformly divide a map into grid maps with equal proportions; obtaining an unmanned aerial vehicle cooperative path planning configuration space according to the simulation environment grid map; and searching in the unmanned aerial vehicle cooperative path planning configuration space by using an unmanned aerial vehicle cooperative path planning algorithm based on a WM-MI-CS algorithm, so as to obtain a feasible unmanned aerial vehicle cooperative path. According to an example task requirement, the multi-index dynamic priority unmanned aerial vehicle cooperative path planning method has the advantages of low operation cost, short task completion time and fast algorithm speed.

This invention relates to the field of underground mine positioning and navigation technology, solving the technical problem of difficult underground positioning and mapping in unstructured environments where GPS is ineffective. Specifically, it relates to a SLAM method for unstructured features in underground mines, comprising the following steps: acquiring point cloud information of the current laser frame from a lidar system; calculating the curvature of each point in the point cloud information of the current lidar frame, and extracting corner features and planar point features of each point based on the curvature; extracting information from the current frame using a camera, detecting corner points of the current frame using the FAST algorithm based on the current frame information, and determining whether it is a corner feature point; and tracking feature points of the current sliding window keyframe using the KLT optical flow tracking algorithm based on the current frame information. This invention achieves high-precision positioning in coal mines, balancing accuracy and computational load, and improving the accuracy and real-time performance of positioning and mapping in mine tunnels.

The invention provides a time domain finite element fast algorithm for electromagnetic force coupling simulation of a loudspeaker, which comprises the following steps of: performing three-dimensional modeling on a micro loudspeaker, setting parameter information of a solving region, and extracting a calculation model; performing discretization processing on the micro loudspeaker model by using a tetrahedral mesh; an electromagnetic force coupling discrete process is deduced, a hybrid finite element fast time domain solver is constructed, a discrete model and solving parameters of the miniature loudspeaker are imported into the solver, and the solver is initialized; and carrying out coil geometric analysis, calculating the current density in the coil, taking the current density as an excitation source of a magnetic field, calling a hybrid finite element fast time domain solver, calculating to obtain the displacement of the vibrating diaphragm, and finally obtaining the time domain vibration curve of the vibrating diaphragm. According to the time domain hybrid finite element fast algorithm, the tree-cotree technology and the Lanczos modal order reduction technology are introduced, the method can be used for analyzing the time domain electromagnetic force coupling simulation problem of the complex micro loudspeaker, and the method has the advantage of being high in solving efficiency under the condition that the solving precision is met.

The present application discloses a method for rapid splicing of three-dimensional surface morphology based on an improved ORB, comprising: determining the overlapping area of ​​the micro-nano surface structure to be spliced ​​based on a block matching method; extracting an initial feature point set based on the FAST algorithm, calculating the Harris corner response value of the feature points after non-maximum suppression processing, and screening according to the response value; describing multiple feature points to be matched with the largest response value to obtain feature descriptors; obtaining coarse matching point pairs based on the feature descriptors of the feature points to be matched, performing fine matching on the coarse matching point pairs using a decoupled RANSAC algorithm, obtaining a three-dimensional space rigid body transformation model, and performing splicing and fusion of micro-nano surface structures through transformation interpolation and overlapping area data fusion. The present application also discloses a device for rapid splicing of three-dimensional surface morphology based on an improved ORB, corresponding equipment, and storage medium. The present application can significantly improve splicing efficiency and accuracy, and effectively solve the problem of scale differences in micro-nano surface structure data in different directions.

The invention discloses an inspection robot dog scheduling method considering charging planning in an electric energy meter detection scene, and belongs to the technical field of inspection scheduling. The technical scheme of the invention comprises the following steps: deeply fusing inspection task allocation and charging planning decision, accurately calculating the shortest feasible path of a robot dog between equipment points and between the equipment points and charging piles, and calculating the optimal path of the robot dog; by means of the technical scheme, the inspection task execution efficiency of the inspection robot dog is improved, the practicability and reliability of the inspection robot dog scheduling scheme are improved, the algorithm response speed is high, dynamic events such as newly-added inspection tasks and robot dog state changes can be flexibly processed, and the service life of the inspection robot dog is prolonged. Rapid replanning of the inspection task sequence is realized, and efficient and stable execution of the inspection task is ensured.

The present invention provides a signal detection method based on Kendall's harmony coefficient, comprising the following steps: obtaining array signals received by multi-channel sensors; calculating the rank statistics of the array signals by row to obtain a rank matrix; calculating the Euclidean square distance of the rank statistics of any two rows of the rank matrix to obtain a distance matrix; calculating the sum of the non-diagonal elements of the distance matrix, and performing linear operations on the summed result to obtain the expression of the test statistic; calculating the corresponding detection threshold according to a given false alarm probability; comparing the size of the test statistic and the detection threshold, so as to determine whether the source signal exists. The method of the present invention not only exhibits robustness against impulse noise, but also has a fast algorithm for the number of channels, and can be applied to fast signal detection in an impulse noise environment.

The invention discloses a 4D wireless spectrum map reconstruction method in a dynamic environment, and the method comprises the steps: carrying out the modeling of missing spectrum data and mixed noise into a four-dimensional tensor model according to the four dimensions of two-dimensional space, frequency and time; in order to reduce the degree of freedom of the tensor model, the model is decoupled into an incomplete spectrum map corresponding to a single radiation source through a non-negative matrix factorization algorithm. Further considering the problem of mixed noise influence and calculation complexity, proposing an optimization function based on a dynamic threshold value and a fast algorithm combining random singular value decomposition and Nesterov accelerated gradient, and solving under a non-negative matrix factorization algorithm framework; and finally, reconstructing the frequency spectrum map corresponding to each radiation source by adopting parallel computing to obtain a complete 4D frequency spectrum map. According to the method, an accurate 4D spectrum map can be obtained based on spatial sparse sampling in a dynamic time-varying environment, and the high-precision spectrum situation sensing requirement in a complex noise environment is met.

The present application discloses a tire polishing curve fitting method, including the following specific steps: scanning data is subjected to mean linear filtering, and coordinate system conversion and processing, the scanning data is converted into polar coordinate representation, the pattern groove data is extracted, and further converted back to rectangular coordinates for maximum filtering to obtain approximate data of the tire surface without the pattern groove; on the basis of data preprocessing, the present application uses logical analysis method to summarize the law of traditional genetic algorithm, proposes an optimal state genetic algorithm, and applies it to tire laser scanning data processing, the optimal state genetic algorithm has the advantages of fast algorithm convergence speed, small calculation amount and the like in processing tire scanning data and fitting tire pattern groove curve, effectively improves the algorithm search speed and accuracy, and makes up for the deficiency of tire scanning algorithm theoretical research and practical application.