28 results about "Robust regression" patented technology

The invention relates to the technical field of radar remote sensing, discloses a high-order loose body historical activity detection method based on multi-source DEM data, and aims to solve the problem that the historical activity of a high-order loose body in a glacier coverage area is difficult to accurately and quantitatively evaluate in the prior art. The scheme mainly comprises the steps that drainage basin units are automatically divided based on DEM data, and glacier influence areas are screened; multi-source DEM data are fused, and system errors are eliminated through geographic positioning deviation correction, elevation distortion deviation correction and track mode deviation correction; establishing a penetration depth physical model, and correcting the radar penetration depth in the glacier area by using the C / X wave band elevation difference optimization parameter; calculating an elevation change rate by adopting robust regression; and calculating an activity comprehensive index by integrating multiple factors, and identifying a hot spot region and a time period through probability estimation and spatio-temporal clustering. According to the method, automatic and high-precision quantitative evaluation on the historical activity of the high-position loose body is realized, and the method is particularly suitable for geological disaster early warning and risk evaluation in high mountain areas.

The invention discloses a new energy box-type substation state abnormity diagnosis method and system, and relates to the technical field of power equipment monitoring. According to the method and the system, through event-triggered high sampling and cross-channel time base self-inspection, multi-source sequence synchronization is ensured, and a consistent time axis is provided for subsequent estimation, so that online analysis under the condition of no power outage is supported; introducing dual-path estimation of RL step fitting and voltage drop robust regression in a transient stage, cooperatively inhibiting a closing peak and outlier data, and obtaining a stable initial value of the contact resistance; in a steady-state stage, adopting Goertzel frequency selection to extract a fundamental wave phasor and performing ohmic purity judgment, and expanding to a resistance-inductance joint model when necessary to remove interference of parasitic inductive reactance on resistance estimation; and then mutual identification and residual constraint are carried out on the temperature and electric power coupling relationship by using a discrete heat RC model, and local write-back correction is carried out on the deviation, so that the consistent closed loop of the electric domain and the heat domain is realized.

The invention provides an ARC test temperature rise rate calculation method and device, and belongs to the technical field of ARC test. The ARC test temperature rise rate calculation method comprises the following steps: S1, obtaining original time temperature data of an ARC test search stage; s2, smoothing the original time and temperature data based on a non-parametric smoothing method to obtain a central trend curve, and scanning the central trend curve based on a change point detection algorithm to extract a plurality of data segments; s3, fitting all the data segments based on a robust regression algorithm to obtain a target linear model; and S4, obtaining the slope of the target linear model, and taking the slope as the temperature rise rate of the ARC test search stage. According to the method, the technical problem that the calculation result is inaccurate and unstable due to data noise and abnormal values when the temperature rise rate at the Search stage of the ARC test is calculated by adopting an end point method in the prior art is solved, the accuracy is improved, the robustness and the repeatability are relatively high, the hardware dependence is reduced, and automatic and intelligent analysis is realized.

The application relates to an input data quality evaluation method for photolithography process window analysis, and the method comprises the following steps: S1, obtaining standardized data; S2, obtaining a robust regression residual analysis confidence degree by using a RANSAC regression algorithm; S3, obtaining an isolation tree confidence degree based on an isolation tree anomaly score; S4, obtaining a local outlier factor confidence degree by using a local outlier factor method; S5, comprehensively obtaining an unsupervised confidence degree based on the isolation tree confidence degree and the local outlier factor confidence degree; S6, obtaining a Gaussian confidence degree; S7, comprehensively obtaining a comprehensive confidence degree based on the robust regression residual analysis confidence degree, the unsupervised confidence degree and the Gaussian confidence degree, and obtaining an input data quality evaluation result for photolithography process window analysis. Compared with the prior art, the application has the advantages of improving the input data quality of photolithography process window analysis.

This invention provides a method and system for assembling a genome stack, a method and system for assembling a genome architecture, and a method and system for assembling a genome sequence, belonging to the field of bioinformatics. This invention optimizes a global loss function using robust regression techniques to determine the stacking relationships between sequencing sequences, eliminating interference from false positive alignments, thereby obtaining a more accurate assembled stack set and avoiding copy number loss or misassembly in the assembled genome. The estimation of stack arrangement based on regression matrix calculations has a higher tolerance for outliers; unless the number of ambiguous sample information exceeds the number of true samples, a "collapse" will not occur. By using resampling techniques to independently sample, assemble, improve, and evaluate sequencing data multiple times, and then integrating them into the final assembled genome, the uncertainty of the assembly results caused by noise in the sequencing data can be reduced, and the sensitivity of the assembly results to the selection of assembly parameters can be decreased.

The application discloses an engineering interference suppression method, system, device and medium based on double scale factors, and belongs to the technical field of engineering monitoring. The method comprises the following steps: performing feature recognition and confidence screening according to phase sequence data and amplitude sequence data of each monitoring point of an engineering, to obtain an engineering interference point set; inputting the phase sequence data into a robust regression model, so that the robust regression model performs interference suppression on the phase sequence data, weighting the phase sequence data based on a first weight function corresponding to a first scale factor, to obtain first interference suppression data; comparing a target phase difference value of the first interference suppression data with a first threshold value, if the target phase difference value is less than or equal to the first threshold value, outputting the first interference suppression data, if the target phase difference value is greater than the first threshold value, weighting the first interference suppression data based on a second weight function corresponding to a second scale factor, and outputting second interference suppression data. Therefore, by implementing the application, effective suppression of engineering interference can be realized.

The present invention relates to a time series cross-calibration method and system for a synthetic aperture radar. The method comprises: screening a target area for time series cross-calibration; generating SAR image data pairs at each moment within a time range; extracting a median matrix of a data slice at a single moment using a window median method, obtaining a median matrix array, performing robust regression fitting with outlier removal, and generating a radiation calibration coefficient array; processing the radiation calibration coefficient array according to a final reference value of the radiation calibration coefficient, and obtaining a final radiation calibration coefficient. By performing robust regression fitting with outlier removal on the median matrix array, interference from outliers is avoided; calculating the radiation calibration coefficient corresponding to each single moment, using the correlation of the radiation calibration coefficients between different moments as a reference, combining the radiation calibration coefficient value measured at the moment to be calibrated and the radiation calibration coefficient value at the previous moment, and determining the final reference value of the radiation calibration coefficient, thereby improving calibration accuracy.

The invention provides an air conditioner load-temperature relation robust regression clustering method and system, and the method comprises the steps: obtaining and processing the historical data of air conditioner operation power and meteorological temperature, and generating an air conditioner load-temperature sample set; the number of regression relations is determined according to the classification of the air conditioner operation state division in the user-defined time period; constructing a fuzzy regression model fused with a maximum entropy principle by introducing an entropy penalty term and an L1 norm; initializing the fuzzy regression model according to the number of regression relationships and self-defined parameter values; and taking the air conditioner load-temperature sample set as the input of the initialized fuzzy regression model, carrying out iterative optimization, and determining the regression relationship between the air conditioner and the air temperature of the building to be evaluated according to the output result. Compared with a traditional method, the method and the system have higher robustness and generalization ability, and can be suitable for air conditioner load modeling requirements in different areas and under different climate conditions.

The invention discloses a structural member static strength fitting method and system based on heuristic segmented robust regression, a medium and equipment. The method comprises the steps that data preprocessing is conducted, specifically, normalization processing is conducted on data, measurement noise is eliminated, and abnormal values are detected and eliminated; segmentation point selection: setting each particle to represent a candidate segmentation point set, and continuously optimizing the position of a group by using a particle swarm optimization algorithm; evaluating the fitness of each particle by using a CHNN network so as to find an optimal segment point set; segmentation robustness regression: for each segment, obtaining a fitting curve of each segment based on a noise characteristic analysis regularization robust regression method, and splicing the fitting curves to obtain an integral fitting curve; and by adopting a dynamic time warping method, evaluating the similarity of each section of fitting curve, and verifying the rationality of the sections and the effectiveness of the regression model. The method has innovative breakthrough in the aspects of segmentation point optimization, robust regression, data dimension reduction, abnormal value detection, model verification and the like, and the precision, the stability and the calculation efficiency of static strength fitting of the structural part are remarkably improved.

The invention provides a weld surface defect intelligent detection method and equipment, and belongs to the technical field of image analysis and defect detection.The method comprises the steps that a weld surface height map is preprocessed to obtain denoised height data; determining left and right boundary points and left and right boundary lines of the welding seam area through longitudinal sampling, robust regression fitting and residual gradient detection based on the height data, and fitting a left reference straight line of the left base metal area and a right reference straight line of the right base metal area respectively; based on the left-side datum straight line and the right-side datum straight line, dynamically constructing a reference datum height for each point in the welding seam area; and based on a difference value between the height data and the reference datum height, detecting defects on the surface of the welding seam in combination with a preset judgment threshold value and a region strategy. Based on the method, the invention also provides corresponding equipment. According to the method, accurate quantification and recognition of various weld joint surface defects such as surface excess weld metal, non-filling, surface undercut, weld leg size, linear dislocation, surface air holes and splashing are achieved.

A surface reconstruction method and system for phase deflectometry of specular reflective surfaces uses two-dimensional Chebyshev polynomials and their gradient polynomials as a basis. Based on the gradient data obtained from measuring the specular surface, an ill-conditioned linear equation system is modeled. This system of equations is then subjected to robust regression calculations using iterative reweighted least squares to obtain a robust estimate of the two-dimensional Chebyshev polynomials, enabling surface reconstruction of the specular reflective surface under test. This invention significantly improves the accuracy of specular surface reconstruction while reducing the algorithm's sensitivity to noise and outliers, thereby enhancing the robustness of the reconstruction algorithm.

The invention relates to the technical field of physical experiment measurement, and discloses a Young modulus measurement method by utilizing a blade edge sub-pixel level tracking technology, which comprises the following steps of: 1, constructing an experiment device and inputting experiment parameters; step 2, acquiring an image sequence and preprocessing the image sequence; step 3, carrying out pixel equivalent calibration and selecting a region of interest; 4, displacement is extracted through edge detection and robust fitting, specifically, a horizontally-arranged cutting edge is fixed to the position of a lower chuck of the metal wire, a camera is arranged to directly face the edge of the cutting edge, and the focal length is kept fixed. Through a core algorithm combining edge detection, Hough transform and robust regression, sub-pixel-level precise positioning of the edge of the blade is realized, subjective interference caused by manual reading and manual calibration is effectively avoided, the accuracy and stability of a measurement result are remarkably improved, and meanwhile, by means of the suppression effect of the algorithm on noise points, the accuracy and stability of the measurement result are improved. And the resistance to environmental interference in the measurement process is enhanced.

The embodiment of the invention provides an SOH exception determination method and device, equipment and a storage medium. The method comprises the following steps: based on a predetermined target charging fragment set, calculating a first SOH estimated value for each target charging fragment based on a capacity increment method, and calculating a second SOH estimated value based on an ampere-hour integral method to obtain a first SOH estimated value sequence and a second SOH estimated value sequence, carrying out fusion calculation on the first SOH estimated value sequence, the second SOH estimated value sequence and historical SOH data obtained in advance, and carrying out fitting on a slope, attenuated along with time, of a full-life-cycle SOH sequence obtained through fusion calculation by adopting robust regression, and determining whether the SOH of the current battery is abnormal or not based on the attenuation slope obtained by fitting, the SOH estimated value at the current moment in the full life cycle SOH sequence, a preset SOH threshold value and an attenuation rate threshold value. According to the method, errors of a single method are counteracted through bimodal data fusion, the accuracy is improved, and the anti-noise capability is improved.

The invention provides a mobile communication site energy efficiency hierarchical calculation and abnormal constraint grouping evaluation method, which belongs to the technical field of mobile communication, and comprises the following steps: S1, carrying out data acquisition and preprocessing; s2, self-adaptive grouping and same-group mainstream energy consumption identification are carried out; s3, estimating the lower bound of the theoretical minimum energy consumption; s4, calculating the same group deviation degree BEDR and the theoretical limit deviation TEDR; s5, calculating an abnormal fluctuation factor subjected to dedependent variables; s6, threshold judgment, weighted grading and confidence output are carried out, and according to the mobile communication station energy efficiency grading calculation and abnormal constraint grouping evaluation method, transverse comparability is ensured through matching of grouping and mainstream peak values; theoretically reachable lower bound constraint is provided by'quantile lower envelope ', and only depending on empirical coefficients is avoided. Robust regression and MAD gating significantly suppress the influence of abnormal values. Explainable approximate evaluation can still be given when a score-lacking table is timed; and confidence and reason explanation are output, so that closed-loop treatment and investment decision making are facilitated.

The present application relates to the technical field of radar remote sensing, and discloses a high-position loose body historical activity detection method based on multi-source DEM data, aiming to solve the problem that the prior art cannot accurately and quantitatively evaluate the historical activity of high-position loose bodies in a glacier-covered area, and the scheme mainly comprises the following steps: automatically dividing a watershed unit and screening a glacier-affected area based on DEM data; eliminating systematic errors by fusing multi-source DEM data, correcting geographical positioning deviation, correcting height distortion deviation, and correcting orbit mode deviation; establishing a penetration depth physical model, using C / X band height difference optimization parameters to correct the radar penetration depth in the glacier area; calculating the height change rate by using robust regression; calculating the activity comprehensive index by comprehensively considering multiple factors, and identifying hot regions and time periods by probability estimation and spatiotemporal clustering. The present application realizes automatic and high-precision quantitative evaluation of the historical activity of high-position loose bodies, and is particularly suitable for geological disaster warning and risk assessment in high mountain areas.

The application discloses an image needle path extraction algorithm, comprising the following steps: scanning an image containing a needle path, and performing threshold segmentation on the image; performing connected domain analysis to obtain a plurality of connected domains, and screening a target region to be extracted; performing straight line fitting on a voxel point in the target region to obtain a straight line, and extracting an end point of the straight line to complete needle path extraction. The application performs robust regression analysis on a needle path coordinate point set, eliminates errors caused by metal artifacts in the image, and has high precision and high reliability in needle path extraction, and can be applied to most scenes.

The invention discloses an engineering interference suppression method, system and device based on double scale factors and a medium, and belongs to the technical field of engineering monitoring, and the method comprises the steps: carrying out the feature recognition and confidence screening according to the phase sequence data and amplitude sequence data of each monitoring point of an engineering, and obtaining an engineering interference point set; inputting the phase sequence data into a robust regression model to enable the robust regression model to perform interference suppression on the phase sequence data, and weighting the phase sequence data based on a first weighting function corresponding to the first scale factor to obtain first interference suppression data, and comparing a size relationship between a target phase difference value of the first interference suppression data and a first threshold value, if the target phase difference value is less than or equal to the first threshold value, outputting the first interference suppression data, and if the target phase difference value is greater than the first threshold value, weighting the first interference suppression data based on a second weighting function corresponding to a second scale factor, and outputting second interference suppression data. And effective suppression of engineering interference can be realized.

The application discloses an infrared focal plane array attitude measurement image defect pixel repairing method and device, and belongs to the technical field of infrared imaging and image processing, which comprises the following steps: performing linear fitting on each row and each column of an infrared focal plane array attitude measurement image containing a fault defect, and calculating row residual sum of squares and column residual sum of squares; arranging the row residual sum of squares and the column residual sum of squares in descending order, taking the intersection of the row and the column corresponding to the K largest residuals, and generating an initial positioning mask of the defect pixel; for each defect pixel, adaptively disabling the pollution direction in four one-dimensional windows of a horizontal direction, a vertical direction and double diagonal lines, and establishing a multi-direction robust regression model by using the pixel samples which are not polluted, and then extrapolating to obtain a multi-direction prediction value of the defect pixel; and using the average value of the multi-direction prediction value of each defect pixel as the repair gray value of the defect pixel. The application can realize blind diagnosis and high-precision repair of defects under the conditions of a single frame, no training and no external reference.

The present application belongs to the technical field of hydraulic support group control, and particularly relates to a hydraulic support drift angle correction method and system for steeply inclined large-mining-height working face. Anchor points are arranged on the end and chute side, and data such as IMU, stroke / pressure, UWB / laser and conveyor attitude are collected, and a virtual strike reference angle is obtained through robust regression and consistency constraint; the overall deviation is determined according to the strike number and sliding window (same number rate / variance / mean), and the staggered strategy of remainder grouping is adopted to perform continuous small-step shaping, so that the adjacent drift angle deviation RMS is less than or equal to the threshold value, and then the residual deviation is zeroed in the single-frame closed loop, thereby solving the problem that the center line of the conveyor or chute is taken as a single reference datum, and finally leading to the problem of more and more deviation of the drift angle of the whole line support.

The adaptive outlier identification method for dam deformation monitoring data disclosed in this invention mainly involves introducing the Baseline Management System (BMS) method to optimize the key influencing factors in the dam deformation monitoring model. Then, a robust regression model for dam deformation is established using the optimized set of key influencing factors. Combined with the Longest Trigger Score (LTS) estimation method, robust regression analysis is performed on the dam deformation monitoring data. This allows for simultaneous regression modeling, outlier identification, and structural deformation prediction without the need for data preprocessing. This adaptive outlier identification method can adaptively overcome the misleading effect of outliers on regression, enhancing the significance of the regression results and effectively improving the accuracy of data prediction.

The invention discloses a full-tensor magnetotelluric data scoring method and device, a medium and a program product, and the method comprises the steps: firstly obtaining the actual measurement data of a magnetotelluric measurement point, carrying out the response synthesis of the actual measurement data through a one-dimensional anisotropic model and a phase drift function, and obtaining the corresponding synthesis response data, the method comprises the following steps: constructing a robust regression model on the basis of the residual error of actually measured data and synthesized response data, then performing data fitting on the actually measured data by using the robust regression model to obtain a corresponding fitting reference line, and finally, calculating the residual error of the fitting reference line according to the actually measured data of the magnetotelluric measuring point at each frequency under different rotation azimuth angles and the residual error of the fitting reference line. And scoring the magnetotelluric measuring points to obtain a measuring point scoring result. According to the method, objective and automatic magnetotelluric measuring point data scoring is realized, and the requirement of three-dimensional inversion for data evaluation can be effectively met.

The application provides a track static irregularity trend prediction method, system, electronic equipment and storage medium. The method comprises the following steps: performing similarity test on a plurality of acquired historical data by using a dynamic time warping algorithm, and representing the similarity test result by using a path matrix; fitting and correcting the path matrix by using a robust regression method to obtain fitting data, and repeatedly performing similarity test on the fitting data until the fitting data meets a preset condition, and then ending the fitting and correction to obtain a dynamic warping path matrix; rechecking necessary mileages in the path matrix and the dynamic warping path matrix by using a correlation coefficient method; aligning the plurality of historical data when the rechecking requirement is met; and training the aligned plurality of historical data by using a long short-term memory network model with preset hyperparameters. The application can make accurate prediction on small-scale static single data in the track irregularity degradation state within a range of several meters.

The invention provides a converter station background harmonic parameter identification method and device, and the method comprises the steps: building a parameter identification equivalent model according to a built parameter identification equivalent circuit, and solving the parameter identification equivalent model through employing a complex frequency domain robust regression algorithm based on a mixed weight function formed by combining a square loss weight function and an exponential square loss weight function; and obtaining the harmonic voltage and the harmonic impedance of the background harmonic. According to the method, the parameter identification equivalent model is solved based on the mixed weight function, the influence of data fluctuation on converter station alternating current side harmonic parameter identification can be reduced, and the identification accuracy is improved. According to the method, the fluctuation characteristics of the harmonic voltage and the harmonic current in the converter station scene are considered, the multi-frequency coupling effect is considered, and the robustness and the precision of the identification method are improved.

The application provides a weld surface defect intelligent detection method and device, and belongs to the technical field of image analysis and defect detection. The method comprises the following steps: pre-processing a weld surface height map to obtain denoised height data; determining left and right boundary points and lines of a weld region based on the height data through longitudinal sampling, robust regression fitting and residual gradient detection, and fitting a left reference straight line of a left base material region and a right reference straight line of a right base material region respectively; dynamically constructing a reference datum height for each point in the weld region based on the left reference straight line and the right reference straight line; and detecting defects on the weld surface based on the difference between the height data and the reference datum height, and combining a preset judgment threshold and a region strategy. The application also provides a corresponding device. The application realizes accurate quantification and identification of various weld surface defects such as surface excess height, incomplete filling, surface undercut, weld leg size, linear misalignment, surface porosity and spatter.

This invention discloses a temperature compensation algorithm for pressure sensors based on a two-stage robust optimization neural network. In the global exploration stage, the weights and bias parameters of the neural network are encoded into optimization vectors, and the optimization algorithm iteratively searches these vectors in the solution space. In each search iteration, a neural network using robust regression as the loss function calculates the loss value corresponding to each optimization vector, and this loss value is mapped to the fitness of a slime mold individual, guiding the optimization algorithm to find the globally optimal parameter vector. In the local fine-tuning stage, the optimal parameter vector obtained in the global exploration stage is used as the initial parameters of the neural network, and a neural network with the same structure as in the global exploration stage, combined with a gradient optimization algorithm, is used to finely adjust the initial parameters to obtain the final temperature compensation model. This invention can achieve high-precision temperature compensation for pressure sensor measurements and improve the stability and reliability of pressure sensors in various application scenarios.

The invention belongs to the technical field of hydraulic support group control, and particularly relates to a hydraulic support navigation angle deviation rectifying method and system for a steeply-inclined large-mining-height working face. Anchor points are arranged at the end and the chute side, IMU, stroke / pressure, UWB / laser, conveyor attitude and other data are collected, and a virtual trend reference angle is obtained through robust regression and consistency constraint; the method comprises the following steps of: numbering according to the trend, judging the overall deviation by using a sliding window (same number rate / variance / mean value), executing continuous small-step shaping by adopting a remainder grouping staggering strategy, entering a single-frame closed loop to return the residual deviation to zero after the adjacent navigation angle deviation RMS is less than or equal to a threshold value, and solving the problem that the central line of a conveyor or a chute is used as a single reference standard; and finally, the flight angle of the whole-line bracket is adjusted and deviated.

The invention discloses a data processing method and system for a phase difference method coating thickness gauge, and belongs to the technical field of nondestructive testing. The method aims at solving the problems that due to Gaussian random noise, periodic electromagnetic interference and outlier accidental noise, frequency-phase difference data are scattered, the linear fitting degree is low, and measurement repeatability is poor. The method comprises the following steps: acquiring original frequency-phase difference data; preprocessing is carried out through filtering and adaptive outlier detection; performing core straight line fitting on the preprocessed data by using robust regression algorithms such as random sample consensus (RANSAC), and extracting a stable linear relationship; according to the method, three types of typical noise can be restrained in a targeted mode, after treatment is conducted through the method, the standard deviation of repeated thickness measurement of the same standard part can be reduced by about 60%-80% compared with a direct linear fitting method, the repeatability and reliability of measurement are greatly improved, and an effective solution is provided for accurate and stable measurement of the coating thickness.

The invention discloses an image classification method for resisting data poisoning attack in a federated learning scene. The method comprises the steps that 1, a central server initializes a global image classification model and sends the global image classification model to participants; 2, the participants train a local image classification model and upload the local image classification model to the central server; 3, the central server calculates and uses a repeated median estimation method to fit a robust regression line, calculates the residual error of each parameter and normalizes the residual error; 4, calculating parameter confidence by the central server, and correcting an extreme value; 5, the central server updates the global image classification model; and 6, repeating the steps 1-5 to carry out iteration so as to obtain an optimal global image classification model, and carrying out image classification prediction by using the model. According to the method, the data poisoning attack in federated learning can be resisted, and the precision of the image classification model under the data poisoning attack is ensured, so that the security of federated learning can be improved.