10results about How to "Iterative convergence is fast" patented technology

The invention discloses an array interference SAR error estimation method and device based on microwave vision, and relates to the technical field of microwave radar imaging and array signal processing. The method comprises the following steps: regarding a phase error between array channels as a to-be-estimated parameter for preprocessed SAR (Synthetic Aperture Radar) image sequence data; based on a joint constraint criterion of a minimum mean square error and a minimum entropy, constructing a relationship among the inter-channel phase error, the three-dimensional image feature and the sparse signal; a fast iteration shrinkage threshold sparse recovery algorithm and a self-adaptive momentum optimizer are adopted to solve the relation, and gradient descent type phase error estimation is achieved; and compensating the estimated phase error, and obtaining a high-precision three-dimensional reconstruction result according to the corrected data. According to the method, the calculation cost is reduced, and the quality of array interference SAR three-dimensional imaging is improved.

The present invention discloses an array interferometric SAR error estimation method and device based on microwave vision, and relates to the technical field of microwave radar imaging and array signal processing. The method comprises: for preprocessed SAR image sequence data, the phase error between array channels is regarded as a parameter to be estimated; based on the joint constraint criterion of minimum mean square error and minimum entropy, the relationship between the phase error between channels, three-dimensional image features and sparse signals is constructed; a fast iterative shrinkage threshold sparse recovery algorithm and an adaptive momentum optimizer are used to solve the above relationship and realize the phase error estimation in a gradient descent manner; the estimated phase error is compensated, and a high-precision three-dimensional reconstruction result is obtained based on the corrected data. The present invention reduces the computational cost and improves the quality of three-dimensional imaging of array interferometric SAR.

The invention provides a time slot allocation method and system based on an adaptive weight particle swarm optimization algorithm, and the method comprises the steps: 1, determining service parameters according to the service demands of all network nodes in a data link network; 2, initializing algorithm parameters; step 3, calculating time delay jitter when a single node allocates time slot resources; step 4, calculating an average value of delay jitter of each node when the multi-node time slot resource is allocated; step 5, updating the optimal time slot pbest and the global optimal time slot gbest of the network node; 6, updating the inertia weight W according to an adaptive weight particle swarm optimization algorithm; and 7, updating the speed and the position of the particle according to the inertia weight updated in the step 6. According to the method, the situation that the particle swarm optimization is prone to falling into a local optimal solution due to the fact that the inertia weight is too large or too small in iteration can be avoided, the convergence speed of the last stage of iteration is increased, the algorithm performance is improved, and the optimal time slot allocation scheme of the data chain can be quickly and accurately found.

The application provides a background velocity field reconstruction method and device based on reflected wave waveform inversion gradient preprocessing, a computer readable storage medium and an electronic device. The method comprises the following steps: based on near-offset information, inverse time migration is performed by calculating a Poynting vector, and a reflected wave field is obtained by reverse migration; the gradient of the velocity is calculated by using a reflected wave full waveform inversion technology based on an adaptive error function, the gradient of the velocity is preprocessed by using illumination energy compensation, the gradient of the preprocessed velocity is used to update the background velocity field, and a more accurate background velocity model is obtained. When the method is applied to model inversion, a more accurate background velocity model can be obtained, and the quality of a migration profile is significantly improved.

The application discloses a kind of AC / DC hybrid distribution network power flow calculation method and device considering incomplete LU decomposition preconditioning, and relates to the field of smart grid.The method comprises the following steps: establishing active / reactive control equation based on VSC steady-state equation;Based on Kirchhoff's current law, establish AC / DC node power imbalance equation;Taylor expansion is carried out on the above equation, and high-order terms are ignored, to obtain linearized unified power flow equation set and matrix form Jacobian matrix;Incomplete LU decomposition preconditioning is carried out on Jacobian matrix, and stable double-conjugate gradient method is used to solve, to obtain power flow solution.The application establishes unified linearized power flow model, combines incomplete LU decomposition preconditioning and stable double-conjugate gradient method, effectively reduces the sensitivity to initial value, improves the robustness and convergence speed of complex AC / DC hybrid system power flow calculation.

The present application relates to a kind of transmission line intelligent route selection method based on adaptive resolution grid and improved ant colony algorithm, comprising: after remote sensing image rasterization, adaptive quadtree image segmentation is carried out, and complete adaptive resolution grid map is obtained;Establish hierarchical analysis model, obtain the construction cost of each grid area in transmission line, i.e.grid cost, establish adaptive resolution grid map cost model;Through improved ant colony algorithm, the transmission line path after inflection point correction is obtained.The present application uses improved ant colony algorithm to speed up the iterative convergence of algorithm, and is combined with adaptive resolution grid, the reduction of grid number can significantly improve the efficiency of algorithm;The size of adaptive resolution grid changes according to the change of terrain on remote sensing map, can avoid the problem that the position accuracy of planned route is low;Inflection point correction is carried out to the planned route, reduces the number of corner tower that needs to be built, saves construction cost.

The invention discloses a multi-level LDPC decoding method based on CSK modulation. The method comprises the following steps: performing radio frequency front-end processing on a received navigation radio frequency signal to obtain a digital intermediate frequency signal, performing carrier stripping on the signal, performing CSK demodulation through a matched filter module, transmitting a correlation value after CSK demodulation and corresponding bit data to a decoding module for performing LDPC decoding operation to obtain decoded telegram data, improving multi-level LDPC decoding performance by utilizing mutually orthogonal symbols modulated by CSK and soft decision information output by CSK demodulation, initializing a confidence vector of a received symbol by utilizing a CSK demodulation output result during LDPC decoding, avoiding recalculation of information to be decoded during the initialization process, thereby having smaller calculation complexity and storage complexity, and solving the problem of balancing decoding performance and complexity in the existing multi-level LDPC decoding method.

The invention discloses a 2-3D image registration algorithm for CT and X-Ray of a target area, and the algorithm comprises the following steps: S1, shooting preoperative CT image data, segmenting a rigid structure near the target area, and carrying out the CT preprocessing of an image; s2, X-ray images of c-arm at different angles of the corresponding parts of the patient are shot, and meanwhile, c-arm internal reference and pre-processing images are obtained; s3, registering the rigid structures on the X-ray based on the plurality of rigid structures segmented on the CT, and outputting a plurality of rigid registration parameters; and S4, marking a target area on the CT by a doctor, constructing a model based on the plurality of rigid registration parameters, the X-ray and the CT image, and predicting the position of the target area in the operation under the c-arm coordinate system based on the constructed model. According to the invention, the registration efficiency and the registration precision can be improved; the iteration convergence speed is accelerated, and the algorithm application scene is improved.

The invention discloses a method for identifying parameters of a hyperbolic tangent model of a magnetorheological torsional vibration damper, which comprises the following steps of: firstly, testing the mechanical property of the magnetorheological torsional vibration damper by adopting a servo fatigue motor to obtain a hysteretic curve of the magnetorheological torsional vibration damper, and obtaining a plurality of groups of data by changing frequency and current; the method comprises the following steps of: identifying parameters of a hyperbolic tangent model by utilizing experimental data, firstly solving a stiffness coefficient k according to a physical rule, then solving a viscous damping coefficient c based on mathematical characteristics of the hyperbolic tangent model, finally limiting identification ranges of other parameters according to the mathematical characteristics, performing parameter identification by applying an optimization algorithm, and finally identifying the parameters of the hyperbolic tangent model. And parameter identification is realized by fitting parameter values under different currents. The method has the advantages that sudden change and unsmoothness of the zero point area are eliminated, the recognition precision and the solution quality are improved, the success rate is greatly increased, the calculation time is greatly shortened, and double advantages in efficiency and performance are shown.

The present application relates to the technical field of deep learning, and particularly relates to a new device cold start method for smart home portrait construction, comprising dividing and processing historical record data sets of home devices to obtain input data sets; combining a label classification model and a meta learning model to perform meta training on a training task set to obtain an optimal initialization parameter group; generating a test task set from the input data set and performing meta testing based on the optimal initialization parameter group to obtain a label classification result; and performing cold start based on the label classification result, wherein the model of the present application can learn a group of initialization parameters with good universality from the historical data of other same kind of devices, and can quickly iterate and converge in the case of a small amount of data of a new device, thereby achieving good classification effect. The present application solves the problem that the use records of new devices cannot be correctly labeled and classified in the small sample scene when constructing a user portrait.