34 results about "Ordered statistics" patented technology

The invention relates to the technical field of data processing, and particularly discloses a data analysis method and device fusing a large number rule, equipment and a medium, and the method comprises the steps: determining a minimum convergence amount through hierarchical aggregation; calculating a cumulative sample mean trajectory and carrying out convergence diagnosis; performing noise attenuation weighting based on LLN convergence characteristics; performing distribution and component decomposition under steady moment constraint; performing consistency check and re-extraction robustness of LLN guidance; lLN-constrained model fitting and uncertainty calibration output are carried out; according to the method, based on noise attenuation weighting of LLN convergence characteristics, samples which are close to a steady state obtain greater influence in estimation, and unstable or sparse samples are naturally weakened, so that self-adaptive suppression of heterogeneity and small sample noise is realized, and the robustness of overall estimation is improved; steady moment constraint is introduced into distribution / component decomposition, it can be guaranteed that components obtained through decomposition are consistent with observation convergence characteristics in the aspect of high-order statistics, and component mismatching caused by extreme values or local fluctuation is reduced.

The invention discloses a radar anti-active slice forwarding interference method based on high-order statistics, and the method comprises the steps: S10, carrying out the preprocessing of a radar receiving signal, carrying out the high-order cumulant construction of the radar receiving signal, completing the signal preprocessing, the calculation of a fourth-order cumulant, and the construction of a feature tensor, the high-order cumulant feature tensor of each time window corresponds to the corresponding time domain signal according to the time dimension and serves as a source of subsequent training input; s20, generating a label mask required by supervision training, and constructing an input feature and a supervision label so as to form a data structure in which the input and the label are paired; s30, constructing a MoE deep neural network based on high-order cumulant characteristics; s40, constructing a loss function and an optimization target, and calculating results of all loss items; s50, constructing an end-to-end blind interference suppression framework; s60, performing model training and parameter optimization; and S70, completing interference suppression and target output generation.

The invention relates to the technical field of visual inspection, in particular to an intelligent detection and evaluation system for aero-engine blade repair performance, and the system comprises an aero-engine blade image collection module which is used for obtaining a to-be-detected aero-engine blade repair area image, carrying out the illumination normalization of the image, and generating a repair area subgraph sequence with coordinates. According to the method, the aero-engine blade repair area image is divided into the overlapped sub-areas with the space coordinate indexes, so that the full coverage of detection analysis and the accuracy of defect positioning are ensured, and the contrast ratio, the energy and other second-order statistics are obtained by calculating the gray-level co-occurrence matrix; and in combination with radial and angular energy distribution extracted by two-dimensional fast Fourier transform, a multi-dimensional texture feature set fusing a spatial domain local pixel relationship and a frequency domain periodic structure is constructed, so that the cladding texture state of a repair region can be described more comprehensively, and the identification capability of texture uniformity and directional anomaly is improved.

This invention discloses a monostatic symmetric nested array MIMO system and a near-field positioning method, belonging to the field of monostatic MIMO near-field positioning technology. This invention can effectively improve the positioning performance during near-field estimation. The monostatic symmetric nested array MIMO system proposed in this invention has a symmetrically distributed transmit and receive arrays. Compared to other symmetric MIMO systems, its differential array is a positionally continuous array with more degrees of freedom. With the same physical array aperture, fewer physical array elements are actually used, reducing costs. The near-field positioning method proposed in this invention calculates the fourth-order statistics of the signal and quantizes them to obtain a virtual differential array of the differential array, achieving accurate near-field positioning of sparse MIMO structure arrays. Furthermore, it can be extended to more sparse MIMO symmetric systems.

The invention discloses a radar ranging method and device and a storage medium, and relates to the technical field of radar ranging, and the method comprises the steps: transmitting a detection signal to a target wall; in response to an echo signal of the detection signal, according to the detection signal and the echo signal, calculating a threshold value of each sampling point through a reference unit averaging mode or an ordered statistical mode; screening out candidate target points of which the signal intensity is greater than a threshold value; if the preset number of continuous peak values appear at the candidate target point, improving the threshold value of the back surface echo peak value of the candidate target point; taking a peak value with the earliest arrival time or the highest signal intensity in the processed candidate target points as a target detection point; and calculating a distance parameter between the unmanned aerial vehicle and the target wall according to the target detection point. According to the invention, the technical effect of accurate distance measurement of the target wall surface can be realized.

This invention proposes a multi-target detection method based on entangled photons. First, a pump light is generated using a continuous narrow-linewidth semiconductor laser, which, after polarization modulation and focusing, illuminates a PPKTP crystal. A spontaneous parametric down-conversion process generates polarization-entangled photon pairs, which are then separated into signal and reference light. These are received by a single-photon detector to obtain two time series. Second, coincidence measurements are performed on the time series in different scanning directions within the observation area to obtain a coincidence count matrix. Then, a two-dimensional constant false alarm rate (CFAR) detection algorithm is introduced, using ordered statistics to estimate background noise and adaptively adjust the detection threshold to screen candidate target points. Third, a density clustering algorithm based on correlation strength weighting is employed, combining the correlation strength and spatial distribution of entangled photon pairs to detect different targets and suppress isolated noise. Finally, the coincidence count peak points within each target cluster are selected as the final detection targets, thereby achieving multi-target detection with unknown numbers and randomly distributed locations.

The invention relates to the technical field of image style migration, and provides a diffusion style migration method and system based on latent distribution re-anchoring dynamic injection, and the method comprises the steps: carrying out the second-order statistical alignment of a content hidden variable and a style hidden variable through a potential covariance re-coloring method; automatically positioning an injection starting point of the style according to the edge intensity of the content image, and injecting hidden distribution reanchoring noise into the initialized mixed noise hidden variable to realize hidden distribution reanchoring; the content self-attention features and the style self-attention features are fused through a soft + dynamic style injection module, in the reverse sampling process of the calibrated hidden variables, the fused attention features are gradually injected into an attention layer in the diffusion process at the injection starting point, diffusion denoising is carried out, and a style migration image is generated. According to the method, the image with rich structure details is protected, and the excellent content structure maintaining capability and the accurate style reappearance are realized at the same time.

Methods, apparatus, systems, and articles of manufacture to estimate cardinality through ordered statistics are disclosed. In an example, an apparatus includes processor circuitry to selects a sample dataset from a first reference dataset of media assets and partitions the sample dataset into m mutually exclusive subsets of approximately equal size. The processor circuitry then estimates a ratio of a sample weighted average and empirical cumulative distribution of an approximately largest order statistic from at least one of the m subsets and generates an estimate of a total cardinality of the first reference dataset by multiplying the ratio by approximately m.

The invention belongs to the technical field of engineering construction, and particularly relates to an automatic muck truck work order generation method based on space-time covariance matching, and the method comprises the following steps: S1, constructing an electronic fence path topological matrix, and constraining a legal path transfer rule between electronic fence nodes; s2, defining a Kalman filtering state vector fusing the position, the speed, the load and the path deviation; and S3, performing work order cutting based on load sudden change and path sequence dual verification. According to the method, the problem of work order mistakes and omissions caused by many-to-many path identification, key node misjudgment and work order conflicts in work order statistics can be solved, and the method has a good market application prospect.

The application discloses a kind of based on dynamic weighting joint area CFAR's unmanned vehicle target detection and speed-distance dictionary construction method, it is related to target detection technical field, including: two-dimensional FFT processing is carried out to radar echo signal, constructs distance-Doppler joint spectrum matrix;In distance dimension, detection is carried out using ordered statistics CFAR, in Doppler dimension, detection is carried out using minimum selection CFAR, and respectively generates continuous type confidence score;Through introducing clutter intensity index and target density index, the weight of distance dimension and Doppler dimension is dynamically calculated, and joint confidence index is obtained by weighted fusion;Adopt adaptive dynamic threshold to carry out target decision, finally the detected target unit is mapped as speed-distance pair, constructs distance aggregation dictionary indexed by speed;The present application effectively suppresses false alarm propagation under dense target environment, improves detection robustness, and provides structured speed-distance dictionary data support for target tracking.

This invention discloses a method and system for pontine infarct segmentation and END prediction based on multimodal joint learning. The method includes acquiring and preprocessing multimodal data; constructing a wavelet transform-based feature encoding network to perform multi-scale decomposition and detail preservation of features; constructing a dual-task guided fusion module to align the deep semantics of clinical variables and imaging features and generate task-specific representations; constructing a Mamba-based global feature aggregation module to model sequence dependencies using a state-space model; constructing a multimodal second-order fusion classifier to enhance the clinical-image interaction modeling using second-order statistics; and employing a two-stage joint training strategy for training and prediction, and outputting the prediction results. This invention utilizes the DWT / IWT mechanism to significantly improve the accuracy of capturing small pontine infarct lesions; it achieves explicit interaction between segmentation evidence and prediction signals, significantly improving the segmentation accuracy of small lesions and the reliability of stroke risk assessment.

The invention discloses a bistatic multi-channel synthetic aperture radar (SAR) synchronous signal extraction method and device, and belongs to the technical field of synthetic aperture radar signal processing. According to the method, a bistatic SAR system with a plurality of receiving channels is constructed, and a synchronization signal and a scene echo signal are modeled into a linear mixed signal by using time delay and amplitude differences caused by physical position differences among the channels; a blind source separation algorithm based on high-order statistics is adopted to carry out centralization, whitening and joint diagonalization processing on multichannel received signals, and effective separation of synchronous signals and target echoes is realized. The method does not need to interrupt the imaging process, does not depend on priori knowledge or complex waveform design, remarkably improves the phase synchronization precision and the system flexibility, is suitable for various bistatic SAR systems, and has good practicability and universality.

The application discloses a kind of based on magnetocardiogram's lesion model construction method and system, comprising: acquisition multi-channel magnetocardiogram and preprocessing are obtained preprocessing observation matrix;Based on second-order statistics blind source separation of second-order statistics blind source separation based on multi-delay covariance weighting joint diagonalization, generate interference suppression weight and reconstruction are obtained purification observation matrix;Coupling separation matrix and interference suppression weight generate inversion prior parameter and establish weighted forward model;Under the constraint of weighted forward model, using the improved FOCUSS of connected domain focus to the sparse reconstruction of purification observation matrix is obtained convergent source distribution;Based on convergent source distribution extraction lesion voxel boundary and reconstruction surface grid, combined with energy projection consistency check output three-dimensional lesion model.The application improves lesion positioning accuracy and three-dimensional model construction stability under strong noise and individual difference conditions.

The invention provides a two-dimensional *-shaped sliding window detection method and system, and belongs to the technical field of vehicle-mounted millimeter wave radar signal processing. The method comprises the following steps: firstly, carrying out boundary expansion on an input original two-dimensional data matrix to obtain an expanded data matrix; secondly, constructing a two-dimensional *-shaped sliding window taking the current detection unit as the center on the extended data matrix, and executing hybrid two-dimensional constant false alarm processing on a reference unit set in the sliding window to obtain a background clutter estimation value; then, generating a detection threshold according to the background clutter estimation value, comparing the detection threshold with the amplitude of the current detection unit, and judging a target trace point; and finally, updating the position of the current detection unit according to a preset path and repeating the steps until traversal is completed, and outputting all target trace point information. The spatial distribution of the reference unit is optimized through the design of the sliding window shaped like the Chinese character'mi ', and collaborative optimization of the detection performance and the processing efficiency is realized by combining a mixed processing mechanism of ordered statistics and unit averaging.

This invention discloses a method for tracing the source of generated images based on spatial-frequency cross-domain second-order statistics. This method extracts spatial domain residual anomalous features and frequency domain anomalous features from the generated image, and constructs a cross-domain second-order statistical coupling relationship between them, forming a cross-domain statistical fingerprint for source tracing. The cross-domain statistical fingerprint matrix is ​​obtained by calculating the cross-domain cross-correlation matrix, and then normalized and compressed to obtain a cross-domain second-order statistical feature representation. Based on the similarity matching or distance measurement results between this feature representation and the statistical fingerprint templates of each source category, the source category or confidence score of the generated image is output. This invention can explicitly characterize the coupling relationship between spatial anomalies and frequency anomalies, and still has strong source discrimination ability and good interpretability and generalization performance even when the generation models are highly similar.

The invention requests to protect a coupling signal separation method based on symmetric tensor joint decomposition, and provides a new BSS method, namely TenSOFO, for solving instantaneous blind source separation from the perspective of tensor decomposition. According to the method, a novel efficient alternating direction multiplier method framework is provided, and two symmetrical third-order tensors are allowed to be decomposed at the same time in an individual difference scale INDSCAL format. In addition, the basic relation between the BSS based on the second-order statistic-fourth-order statistic SO-FO and the combined INDSCAL decomposition is established, and the provided INDSCAL method can be effectively applied to the instantaneous BSS task. The TenSOFO has the advantages that memory is saved, effective decomposition is achieved, Gaussian noise items disappear when fourth-order FO is used, and source signals can be processed without prior information of other levels. Therefore, the TenSOFO is particularly beneficial to processing long-time sequence signals such as voice and audio sampled at a high rate. Experimental results show that the TenSOFO method provided by the invention is minimum in error measurement and optimal in performance.

The invention belongs to the technical field of millimeter wave radar non-contact vital sign monitoring, and particularly relates to a non-contact off-bed detection method based on a millimeter wave radar. The method specifically comprises the following steps: firstly, screening a potential radar data window containing a human body target through an adaptive energy-entropy dual gating mechanism; secondly, adopting ordered statistical constant false alarm rate detection OS-CFAR for the screened windows, detecting effective human body targets and bed body targets, and performing further screening in combination with morphological filtering and an adaptive area threshold method; and finally, radar point cloud generation is carried out on the screened targets, radar point cloud only containing human body targets is extracted, and real human body targets and dynamic interference are accurately distinguished through point cloud clustering and space-time consistency joint evaluation.

The application discloses a high-efficiency two-dimensional constant false alarm rate detection method, which comprises the following steps: after the velocity dimension data of the same distance unit is subjected to FFT processing, a threshold is updated; extreme value judgment is carried out on the data greater than the threshold to obtain an extreme value; a reference unit is selected in a distance-Doppler two-dimensional plane with the extreme value as the center to establish a rectangular reference window; the maximum data value in the velocity dimension data of the same distance unit in the rectangular reference window is selected to establish a new reference window; and the extreme value is compared with all the reference units in the new reference window by using an ordered statistical constant false alarm rate (OS_CFAR) algorithm to judge whether the extreme value is a target. The application can not only meet the detection requirements of multiple targets in a complex environment, but also reduce the time consumption in the sorting process during detection, realizes fast search of targets in a large airspace range, and improves the possibility of engineering implementation.

This invention discloses a method for enhancing motor fault data based on a spectral normalization-conditional generative adversarial network (GAN) and high-order statistics for screening, belonging to the field of motor fault diagnosis and data augmentation. The method includes: first, constructing a filter incorporating three aspects: energy efficiency, physical stability, and fault saliency, and a generative adversarial network (GAN) incorporating spectral normalization; second, generating initial fault signal samples using the generator and filtering them through the filter; inputting the filtered fault signal samples and collected real fault signal data into a discriminator for discrimination; and updating the GAN parameters based on the discriminator's discrimination results to obtain a trained generator; finally, using the filter and the trained generator, specifying the fault type, yields new fault signal samples corresponding to that fault type. This invention effectively enhances the diversity of data samples, providing a reference and approach for the practical deployment and optimization of deep learning models in industry.

The invention belongs to the technical field of iterative decoding, and particularly relates to a soft iterative decoding method of a BCH (Broadcast Channel) code, which comprises the following steps of: for the BCH code, expanding and optimizing a standard check matrix of the BCH code to obtain a preprocessed check matrix; performing quasi-BP iterative decoding, performing expansion preprocessing on the input information based on an expansion factor, completing an iterative decoding process, and optimizing and merging weight coefficients; constructing a quasi-BP iterative decoding and ordered statistical decoding OSD hybrid decoding architecture, and when the quasi-BP iterative decoding is judged as an undetectable decoding error UDE, triggering an ordered statistical decoder OSD to perform secondary decoding; a scattered point type external information transfer graph analysis tool is introduced for parameter optimization; and carrying out soft iterative decoding on the subsequently input BCH code, and carrying out parameter optimization. According to the method, a scatter type EXIT tool is introduced to quantify and optimize parameters, the quasi-BP algorithm is universal and parallel, and high reliability, low complexity and low time delay of decoding are considered.

This invention relates to the field of on-orbit telemetry data analysis, and particularly to a method for in-depth analysis of on-orbit data fault diagnosis results based on high-order statistics. First, historical fault diagnosis conclusion data is segmented by a preset time length, and the types, durations, and switching processes of the diagnosis conclusions are statistically analyzed to form high-order statistical information. Second, different conclusion types are labeled with color bands of different lengths according to their duration, and these bands are then stitched together according to the switching process of the diagnosis conclusion types to form a changing color band diagram. Next, based on manually labeled fault tags, a diagnostic model is trained using a deep learning algorithm. Finally, the statistical analysis and color band diagram drawing process are repeated for new data, and the trained diagnostic model outputs abnormal diagnosis conclusions. This invention can effectively filter out false alarms, uncover deep fault patterns, significantly reduce the burden of manual analysis, and achieve keen early warning of early fault symptoms, thus improving the automation level and reliability of on-orbit data fault diagnosis.

This invention discloses a method and system for controlling the wire winding of a power pole frame based on real-time tension feedback. Specifically, it relates to the field of wire winding tension control technology in the manufacturing of prestressed concrete power poles, and is used to solve the problems of tension feedback distortion and uneven prestressing caused by the dynamic changes in the friction coefficient between the steel wire and the frame in existing technologies. By acquiring the steel wire tension feedback value and the axial position information of the winding head in real time, the system judges control anomalies based on tension deviations. When an anomaly occurs, it continuously calculates the higher-order statistics of the tension value and identifies the directional change of the friction state based on its evolution trajectory. It correlates this evolution feature with the axial position to determine the spatial propagation characteristics of the friction state change, evaluates the degree of tension transmission distortion based on this feature, and finally dynamically compensates and adjusts the tension control system parameters according to the degree of distortion. This method can effectively identify and compensate for tension distortion caused by time-varying friction, and improve the uniformity and accuracy of prestressing application.

Methods, apparatus, systems, and articles of manufacture to estimate cardinality through ordered statistics are disclosed. In an example, an apparatus includes processor circuitry to selects a sample dataset from a first reference dataset of media assets and partitions the sample dataset into m mutually exclusive subsets of approximately equal size. The processor circuitry then estimates a ratio of a sample weighted average and empirical cumulative distribution of an approximately largest order statistic from at least one of the m subsets and generates an estimate of a total cardinality of the first reference dataset by multiplying the ratio by approximately m.

The invention provides an uncertainty-guided analytic high-resolution range profile class incremental learning method, and mainly solves the problems that catastrophic forgetting is likely to occur in radar high-resolution range profile target recognition, and old class recognition performance is remarkably degraded. According to the method, multi-resolution structured feature representation is constructed, a statistical prototype containing a mean value and a second-order statistical magnitude is taken as a unique carrier of old category knowledge, variance or covariance information is updated through recursion, and old category distribution characteristics are depicted under the condition of no original sample; on the basis of a recursive least square closed-form solution of recursive tracking of a covariance matrix, progressive analytic updating is carried out on the classifier to replace back propagation training; and a self-adaptive prototype distillation and uncertainty guiding mechanism based on a statistical prototype is introduced, and old knowledge is kept in the incremental training process. Experimental results show that under the condition that old class samples are not used at all, high and stable overall recognition accuracy can be kept in a target class continuous expansion scene.

The application provides a hydrate and gas occurrence form reconstruction and voxel number control method, and belongs to the field of digital rock physics and pore-scale medium reconstruction. The application reconstructs different material occurrence forms in a three-dimensional voxel pore candidate domain, and realizes accurate, stable and repeatable control of saturation in the sense of discrete voxel counting. Specifically, a three-dimensional voxel digital core is taken as input, a pore voxel set or a subset thereof is determined as a candidate domain based on identification of a solid skeleton voxel set and a pore voxel set, a continuous scalar priority field is constructed in the candidate domain, a target saturation is mapped to a target voxel number, and then a target voxel number selection process of "a small amount of pseudo-random disturbance breaking an isovalued platform, a threshold determined by partial order statistics, accurate repair of supplementing / removing, and random seed and rule recording" is performed in the candidate domain to obtain a target phase voxel set.

The application provides a kind of detection method and system for the appearance of pastry, including collecting pastry image input U type detection network to do multi-scale feature extraction, the channel of each level feature map of encoder is normalized and used as the initial feature of jump connection, the texture activity index is constructed by the mean of channel variance, the initial feature of Skip connection after channel normalization is processed by discrete wavelet transform in jump connection, and high and low frequency components are separated;Low frequency is processed by depth separable convolution, high frequency adopts channel second-order statistics weighting, and inverse wavelet reconstruction is obtained after scaling by texture activity index, and the fusion feature map is obtained;The offset field is generated by the convolution of decoder level feature and fusion feature map, the maximum single-axis offset of offset is limited according to texture activity, and the feature position correction is completed;According to the comparison between the maximum response of decoding output probability map and correction threshold, it is determined whether the pastry has appearance defect and the detection result is output.

The application discloses a UAV remote sensing image dark light enhancement method based on manifold-guided RWKV optimization of large model semantic coding, relates to the technical field of remote sensing image processing, and comprises the following steps: constructing a vision-language semantic prior extraction module; constructing a manifold-guided RWKV post-processing module, and constructing a feature extraction network based on the manifold-guided RWKV post-processing module; constructing an enhancement reconstruction decoding module; inputting multi-scale features output by the feature extraction network into a decoder to generate enhanced UAV remote sensing images; training each module based on a joint loss function to generate an optimal model; and performing enhancement processing on input target UAV dark light remote sensing images by using optimal model parameters. The method introduces the second-order statistics of features into sequence calculation, and accurately describes the topological structure changes of ground objects by using a logarithmic Euclidean metric on a symmetric positive definite Riemann manifold, so that the sensitive perception ability of the network to complex ground object texture boundaries is greatly improved.

The application discloses a radar ranging method, equipment and storage medium, and relates to the technical field of radar ranging. The method comprises the following steps: a detection signal is sent to a target wall; in response to an echo signal of the detection signal, a threshold of each sampling point is calculated by a reference unit average method or an ordered statistical method according to the detection signal and the echo signal; a candidate target point with a signal strength greater than the threshold is screened out; if the candidate target point has a preset number of continuous peak values, the threshold of a rear surface echo peak value of the candidate target point is increased; a peak value with the earliest arrival time or the highest signal strength in the processed candidate target point is taken as a target detection point; and a distance parameter between a target wall and an unmanned aerial vehicle is calculated according to the target detection point. The application can achieve the technical effect of accurately measuring the distance of the target wall.