19results about How to "Improve algorithm efficiency" patented technology

The invention discloses a three-dimensional color modeling method and system based on multi-sensor fusion, and the method comprises the steps: obtaining monocular vision pose information, an RGB image and radar three-dimensional point cloud information, enabling the monocular vision pose information to be used for correcting a radar pose, and generating corrected point cloud information; the corrected point cloud information is screened and then projected to a specific plane; proper feature points are extracted from the projection image and the RGB image, and according to the corresponding relation between the feature points of the projection image and the RGB image, the plane point cloud of the projection image and the pixel coordinate plane of the RGB image are registered; performing fusion modeling on the point cloud and the image data by using an MRF model; according to the method, the radar pose is corrected by using visual pose information, the accumulated error influence of a system is reduced, the corrected point cloud is projected to a plane, and feature points of a projection image and a color image are extracted to establish an MRF model for describing the relationship between a pixel plane and the corresponding point cloud, so that a point cloud image which is uniformly distributed in space and contains color information is obtained.

The invention relates to a power data federated learning privacy protection method based on efficient homomorphic encryption. The method comprises the following steps: step 1, a client initializes a neural network model; step 2, the server side encrypts and updates local neural network model gradient parameters uploaded by the client side; 3, the server side executes gradient average aggregation operation, the server side does not need to decrypt, the encrypted gradient parameters uploaded by the client side are directly calculated, an average value is solved, and therefore global gradient parameters are updated; step 4, after each client receives the updated global gradient, carrying out decryption calculation operation; 5, loading the updated weight parameters of the neural network model into a local neural network model, and performing neural network model training by using a local data set; the problems that an existing encryption technology is large in operand, large in computing resource overhead and low in algorithm efficiency, and privacy leakage exists in the transmission and analysis process of power data can be solved.

The invention provides a rotating machine fault diagnosis method based on OCSSA-VMD-CNN-BILSTM. The rotating machine fault diagnosis method comprises the following steps: inputting a rotating machine vibration signal; obtaining a range of a vmd decomposition layer number K value through a center frequency method, and setting a range of a penalty factor alpha; and taking the minimum sample entropy as a fitness function, and performing parameter optimization on the VMD by using an improved sparrow optimization algorithm OCSSA to obtain a [K, alpha] optimal combination. According to the method, the bearing fault diagnosis model is built based on the CNN-BILSTM model, the multi-strategy fusion improved sparrow search algorithm (OCSSA) is utilized to optimize the VMD parameters, and when the model is subjected to hyper-parameter optimization, the OCSSA has extremely high convergence speed and extremely high test precision. According to the method, the optimal parameters of the vmd can be obtained without priori knowledge, the algorithm efficiency is improved, manual parameter screening is avoided, higher diagnosis precision and stability are achieved, and the generalization and intelligent level of a diagnosis model are greatly enhanced.

The invention provides a satellite constellation simulation method based on a native set model. The satellite constellation simulation method comprises the following steps that a current view angle rotation matrix is obtained through calculation according to the current moment; and calculating a satellite orbit. At the T0 moment of each circle, calculating satellite coordinate points contained at the starting and ending moments of the current circle to form a satellite orbit, calculating the satellite coordinate positions by introducing a front-end library, when the next circle arrives, calculating the satellite orbit of the current circle again at the T0 moment of the circle, and updating and displaying the satellite orbit; introducing a front-end library to calculate a satellite coordinate position; the satellite orbit and the view angle keep the same view angle rotation matrix for rotation transformation. According to the method, the algorithm efficiency is greatly improved, the interaction process of the front end and the rear end is saved, the calculation and display efficiency of the satellite orbit is improved and the rendering performance is improved on the premise that the display effect of satellite simulation is not influenced, and the native model supports dynamic updating, so that the geometrical shape and the material can be modified during operation, and re-creation is not needed.

One or more embodiments of the present specification provide a color lookup table generation method and a color correction method, which divides the color range into multiple color intervals, and partitions each color interval to calculate the linearity, and adaptively adjusts the sampling step size of each color interval based on the linearity, effectively balancing the contradiction between too many sampling points being time-consuming and too few sampling points being insufficiently accurate, and adaptively allocating the sampling point distribution of each color interval, so that the sampling points are more sparse in areas with higher linearity, saving sampling time and improving algorithm efficiency, while the sampling points are denser in areas with lower linearity, reducing the interpolation error in the color correction process and improving the color correction accuracy.

The invention discloses a satellite image compression method based on linear frequency modulation, and the method comprises the steps: S1, obtaining a satellite image, and carrying out the partitioning of the satellite image through employing a self-adaptive partitioning method, so as to enable different projective transformations in the satellite image to be separated, and obtaining a plurality of image blocks; s2, inputting all the image blocks into a trained neural network model, and predicting to obtain chirp transformation parameters of each image block; s3, performing discrete chirp transformation on each image block according to the chirp transformation parameter corresponding to each image block so as to realize transformation from an image domain to a chirp frequency spectrum domain; s4, in the chirp frequency spectrum domain, the amplitude of the frequency component is quantized, so that continuous frequency spectrum coefficients are mapped into a discrete set; and S5, compressing quantization results of all the image blocks in the step S4 by adopting a lossless algorithm, and then encoding the compressed spectrum data by adopting an encoder to complete compression of the satellite image.

The present application belongs to the technical field of ground laser radar forestry data processing, and discloses a broad-leaved forest single tree segmentation method and system based on branch information guidance, taking ground-based laser radar broad-leaved forest point cloud as a processing object, using RANSAC cylindrical fitting to combine the growth characteristics of the tree trunk to detect the tree trunk; starting from the top of the tree trunk in the low vegetation area, the tree branches are extracted by segmenting and growing the branches and combining the thickness changes of the branch segments; starting from the end of the branch, the tree crown leaf point cloud is segmented by layer-by-layer growth. The present application has stronger trunk detection capability under the conditions of lush low vegetation and complex terrain; on the other hand, the present application can accurately segment the tree crown when the large and small crowns are intertwined and the multiple crowns are closely surrounded. The algorithm is efficient, simple and easy to use, and has great practical significance for improving the semantic understanding ability of forest scenes, assisting forest resource investigation, vegetation ecological research, and satellite remote sensing product calibration and verification.

The invention relates to the technical field of digital signal processing in optical communication, in particular to an FPGA (field programmable gate array) of parallel FFT (fast Fourier transform), electronic equipment and a design method.The FPGA of parallel FFT comprises one or more serial modules, each serial module comprises a plurality of serial sub-modules, and each serial sub-module comprises one or more serial sub-modules; each serial sub-module performs butterfly operation on fast Fourier transform (FFT) data of FFT to obtain a first operation result; the parallel module comprises one or more parallel sub-modules, a first operation result output by each serial sub-module of the serial module is input to the input end of the corresponding parallel sub-module, and the parallel sub-module carries out parallel processing of butterfly operation on each first operation result and outputs second operation results of corresponding paths; and converting the FFT data into data of corresponding paths. Therefore, the problems of low working efficiency of the IP core of the FPGA, relatively low adaptability to high-performance applications, relatively large resource consumption and the like in related technologies are solved.

The present invention discloses an image target extraction method based on LRW and density clustering, comprising the following steps: S1, LRW superpixel algorithm pre-segmentation; S2, superpixel texture feature extraction; S3, density similarity measurement: constructing a multi-feature similarity matrix M between different superpixels ij ; S4, select parameters ε and Minpts; S5, complete image target extraction. The goal of the present invention is to extract the image target based on the differences in texture, color and structure, and use a clustering method to effectively combine the differences between the various microscopic components to achieve color image segmentation.

The disclosure provides a social network data analysis method and system based on limited k-means, and relates to the technical field of social network data processing. In the method, the initialization center selection stage mainly considers the connected constraint. After the first center is randomly selected, the remaining clustering centers are selected by cyclically calculating the weight probability affected by the connected constraint. Due to the influence of the connected constraint, the centroid of each connected set is used to represent the data points in the connected set to solve the limited k-means problem. Then, in the assignment step of the algorithm iteration stage, for the two types of data constraints, the strategy of preferentially processing the disjoint non-connected set and preferentially considering the intersection of the disjoint non-connected set and the connected set is adopted, and the constraint points are classified and processed, so that higher algorithm efficiency is achieved. The disclosure solves the problem of inaccurate data processing of the connected constraint and the disjoint non-connected constraint in the clustering process.

The invention discloses a unique primary key identifier generation method and device, and the method comprises the steps: reading a to-be-processed data file, obtaining a total number value of the file record number of the to-be-processed data file, generating non-repeated initial random numbers with the number corresponding to the total number value according to a preset script, and generating a unique primary key identifier of the to-be-processed data file; and storing the non-repeated initial random numbers with the number corresponding to the total number value into a list container, for each file record in the to-be-processed data file, obtaining the date of the primary key identifier and the batch number of the primary key identifier from the file record, and reading a target random number from the list container according to a preset reading mode, the preset reading mode is used for ensuring the uniqueness of the read target random number, and generating a unique primary key identifier recorded by the file according to the date of the primary key identifier, the batch number of the primary key identifier and the target random number.

The embodiment of the invention discloses a meta-information extraction method and device, a server and a storage medium. The method comprises the following steps: acquiring operation data and initial meta-information corresponding to the operation data; analyzing the operation data to obtain output meta-information corresponding to the operation data, the output meta-information at least comprising data attribute information, data processing information and format information of a video included in the operation data; and extracting target meta-information of a preset type from the initial meta-information, and constructing a data packet containing the output meta-information and the target meta-information as meta-information corresponding to the operation data. By applying the scheme provided by the embodiment of the invention, when a user needs to operate the meta-information corresponding to the data, the meta-information can be directly downloaded from the server without downloading the whole operation data and then analyzing the whole operation data to obtain the required meta-information, so that the meta-information obtaining speed of the user can be improved, and the overall algorithm efficiency of the user is improved.

The embodiment of the present application provides a method for calculating respiratory frequency and its device, and a smart wearable device, the method for calculating respiratory frequency includes: obtaining a first PPG signal of a user's pulse collected by a PPG sensor; performing low-pass filtering and baseline drift processing on the first PPG signal to obtain a second PPG signal; using a signal decomposition algorithm to process the second PPG signal, decomposing a first intrinsic mode function signal whose frequency is less than the minimum preset frequency value, and calculating a mixed signal including a respiratory signal and a residual signal based on the intrinsic mode function signal; and calculating the respiratory frequency based on the mixed signal. This method no longer decomposes the entire second PPG signal, but only needs to remove the high-frequency signal in the second PPG signal, and calculate the mixed signal including the respiratory signal and the residual signal based on the retained low-frequency signal and the residual signal, and then completes the calculation of the respiratory frequency based on the mixed signal, optimizes the process of the signal decomposition algorithm, and improves the efficiency of the algorithm for calculating the respiratory frequency.

This invention relates to the field of robot skill learning technology, specifically to an efficient robot skill learning method, system, device, program product, and medium. The method includes: obtaining the robot's task policy equation in Cartesian space; obtaining the expected trajectory at each time step of the robot's task policy; constructing the cost function of the robot's task policy; and obtaining the robot's final expected trajectory. This invention utilizes interactive reinforcement learning to integrate human and machine intelligence, improving algorithm efficiency and overcoming the limitations of existing reinforcement learning methods in robot operation applications. It enables real-world robot skill learning, allowing robots to be applied in a wider range of scenarios.

The present application relates to the technical field of wireless communication, and particularly relates to a kind of super-diagonal RIS enhanced multi-target sensing integration optimization method and system. Including: establishing super-diagonal RIS enhanced multi-target sensing integration system model;Under the constraints such as meeting user minimum rate constraint and base station transmitting power constraint, with base station transmitting beamforming matrix and super-diagonal RIS scattering matrix as optimization variable, the sensing estimation Cramer-Rao bound minimization is constructed Sensing integration joint optimization problem;The joint optimization problem is solved by alternating optimization algorithm, on the one hand, the Shur complement convex conversion technology and semi-positive definite relaxation method are used to optimize base station transmitting beamforming matrix, on the other hand, the adaptive Riemann steepest descent method based on geodesic is used to optimize super-diagonal RIS scattering matrix under unitary constraint, and the common promotion of sensing performance of super-diagonal RIS enhanced multi-target sensing integration system is realized by the alternating optimization of two parts.

The invention discloses an electric vehicle charging optimization method, system and device and a readable storage medium. The method comprises the steps that the starting charging time and the ending charging time of the day are determined; when the initial charging time is reached, monitoring whether an electric vehicle is accessed to be charged, and if so, continuing the subsequent steps; judging whether the accessed electric vehicle urgently uses electricity, if so, adopting high-power quick charging, and otherwise, adopting low-power slow charging; solving the minimum value of the charging target function by adopting an adaptive genetic algorithm in an iterative optimization manner, and obtaining an optimized charging strategy; and controlling the connected electric vehicle to be charged orderly, continuously monitoring the charging condition of the electric vehicle and whether a new vehicle is connected or not, and when the new vehicle is connected, carrying out demand judgment again and updating the optimal charging strategy. According to the method, the charging economy of the electric vehicle user and the operation safety of the power distribution network are considered, the charging target function is solved by using the adaptive genetic algorithm to obtain the optimized charging strategy, and the operation stability of the power grid is ensured.

The invention discloses a chip layout method based on area minimization. The method comprises the following steps: converting an area minimization problem into a two-dimensional rectangular strip boxing problem; inputting module information and container information; the candidate length Lh of the module information is preliminarily screened by calculating the linear combination of the lengths of the module information, and the candidate length Lh is not smaller than the maximum width of the module information; establishing a one-dimensional continuous boxing problem model by the group of candidate lengths Lh, and solving to obtain a lower bound of the group of candidate lengths Lh and a solution set of the positions of the group of modules along the Y axis; screening the candidate length Lh again to obtain a minimum candidate length Lh1, and detecting the legality of the x value of the module information; detecting whether the result is qualified or not by adopting an x-check algorithm; according to the chip layout method based on area minimization, the two-dimensional module arrangement problem is relaxed into the one-dimensional module arrangement problem, and a high-efficiency accurate solving algorithm is provided for the area minimization problem in the chip layout design.

The invention discloses a method for path planning of an unmanned ship on water, and relates to the field of path planning. The method for path planning of an unmanned ship on water comprises the following steps: constructing a mathematical model of the motion of an unmanned ship on water surface; analyzing the optimization target of the unmanned ship on water surface in path planning; obtaining the dynamic and kinematic models of the unmanned ship according to the established mathematical model, and describing the motion law of the ship; studying the autonomous obstacle avoidance strategy of the local area of ​​the unmanned ship on water surface; studying the optimal control of the global path planning of the unmanned ship on water surface; global path planning refers to the path planning made after fully understanding the surrounding relevant environment. The global planning needs to master all the environmental information in the motion space, and arrange the path planning according to this information. Study a variety of global path planning methods, seek the optimal control algorithm, and finally perform simulation verification on the above obstacle avoidance strategy and path planning algorithm.

The invention discloses a nuclear magnetic resonance pure chemical shift data optimization method and device based on deep learning, and the method comprises the steps: constructing a nuclear magnetic resonance pure chemical shift data optimization model, first training data and second training data, carrying out the pre-training of the nuclear magnetic resonance pure chemical shift data optimization model through employing the first training data, and carrying out the pre-training of the second training data; a pre-trained nuclear magnetic resonance pure chemical shift data optimization model is obtained, the pre-trained nuclear magnetic resonance pure chemical shift data optimization model is finely adjusted through second training data, and a finely-adjusted nuclear magnetic resonance pure chemical shift data optimization model is obtained; and extracting an FID signal in the nuclear magnetic resonance pure chemical shift data before optimization and performing Fourier transform to obtain a corresponding Fourier transform result, and inputting the fine-tuned nuclear magnetic resonance pure chemical shift data optimization model to obtain a frequency domain signal of the optimized nuclear magnetic resonance pure chemical shift data. The problems that an existing deep learning model is insufficient in generalization and low in precision are solved.