50 results about "Robust regression" patented technology

A measurement system and a measurement method, which can obtain a measurement value close to a true value considering an overlay measurement error according to a higher order regression analysis model. The measurement system and the measurement method provide a technique for determining optimal positions of shots to be measured using an optimal experimental design. When the regression analysis model and the number of shots to be measured are determined in advance, a method is used for determining an optimal number of shots to be measured according to the regression analysis model and process dispersion using a confidence interval estimating method. A dynamic sampling method is used for dynamically changing the number and positions of shots to be measured according to a change in process features by combining the above two methods. And, when erroneous data is detected, or when measured data is missing, a robust regression analysis method and a technique for filtering the erroneous data and the missing data are used.

The invention relates to a radio frequency map self-adaption positioning method based on a clustering mechanism and robust regression. According to the method, in the offline training stage, signal features of multiple reference points and check nodes of one scene are collected firstly, and a static radio frequency map is established; then, a path loss parameter of each reference point is calculated according to the static radio frequency map of the scene; the reference points are clustered according to the path loss parameters, and a positioning area is divided into a plurality of subareas; finally, RSSI (received signal strength identification) fingerprints of a reference point in each subarea and RSSI fingerprints of a check point in the subarea are subjected to linear robust regression; and in the online positioning stage, the radio frequency map is updated by check point RSSI vectors and robust regression parameters which are acquired at regular time, and then, the radio frequency map is used in a weighted K neighbor algorithm, so that the positioning is realized. The method is simple, easy to implement and high in positioning accuracy, influences of outdating of the radio frequency map on positioning calculation can be effectively reduced, and the outdating of the radio frequency map is caused by factors such as RSSI random jittering, interference of walking of indoor workers and the like.

The invention relates to a forecasting method of baking sheet smoke, in particular to the baking sheet smoke forecasting method based on robust regression modeling. A model from physicochemical index object to a smoke index object is built through the existing baking sheet physicochemical data and smoke data, and a baking sheet smoke value of unknown baking sheet smoke data can be directly forecasted by using the physicochemical ingredient data. By means of a robust regression model, malpractices caused by singular value samples in the physicochemical data and the smoke data can be effectively avoided, robustness of the model can be guaranteed to great extent, the system can effectively forecast the smoke value of the baking sheet, and whole quality situations of the baking sheet can be well estimated.

The invention discloses a signal intersection single vehicle stopping delay time estimating method based on GPS data. The method comprises steps that: firstly, vehicle GPS data of vehicles at a signal intersection is firstly acquired; secondly, whether the vehicles are in a saturated state is determined according to stopping positions; single vehicle stopping delay time calculating models in saturated and unsaturated states are respectively established; and single vehicle stopping delay time in the saturated and unsaturated states is acquired. The method utilizes the linear relations between the signal intersection single vehicle stopping delay time and the stopping positions, on the basis of considering about pollution to delay data caused by GPS data abnormality and accidental vehicle stopping, the relation models between the single vehicle stopping delay time and the stopping positions in unsaturated and over-saturated states are respectively established by utilizing the robust regression thought and employing least trimmed squares LTS regression to estimate stopping delay time of single vehicles at the signal intersection; tracking and positioning for the single vehicles are realized, and the method has strong timeliness.

The invention relates to a method and device for leakage detection and leakage location in an area of a supply network (e.g. water supply, gas supply or district heating network), wherein measurement values of sensors of the supply network are statistically analysed for the presence of leakages using robust regression methods. The false alarm rate (type 2 error) is in particular minimised.

An intersected radiometric calibration method for a satellite-borne multispectral infrared sensor comprises steps of S1, selecting out the optimal intersected radiometric calibration area of the satellite sensor according to REF data files and MON data files; S2, evenly dividing the optimal intersected radiometric calibration area into space matching grids at intervals with the same longitude and latitude; S3, establishing a space-time statistical lookup table, wherein the space-time statistical lookup table is formed by a space index table, a time index table and a file statistical table, and space matching grid aggregation information is written in the file statistical table; S4, selecting out radiance matching data meeting matching conditions in the space matching grid aggregation information according to the space matching grid aggregation information; S5, establishing statistical regression relationships among all the radiance matching data by means of a linear robust regression method to estimate intersected radiometric calibration coefficients.

The invention discloses a method for evaluating the development coordination of an island microgrid. The method comprises the following contents: 1) analysing from the four aspects of distributed power side, load side, national economy and environment of the island microgrid to construct a development coordination evaluation index system for the island microgrid; 2) carrying developed illustration on coordination indexes with island microgrid characteristics, and establishing a corresponding calculation model; 3) improving a rough set theory by virtue of a multi-interval partition method; 4) determining an index weight matrix by virtue of the improved rough set theory; 5) fusing the weight matrix by virtue of a robust regression algorithm, and determining index weight values; 6) analysing the influence of each index in the island microgrid and the conditions which need to be met during planning and construction according to each index weight, and outputting a development coordination evaluation report for the island microgrid. The evaluation system and evaluation method disclosed by the invention are capable of scientifically and objectively evaluating the development coordination of island microgrids with different structures. Therefore, scientific guidance basis can be provided for the planning and construction for the island microgrids.

The disclosure provides an approach for denoising (also referred to as “filtering”) rendered images. In one embodiment, a denoising application takes as input rendered images and feature buffers that encode image information such as surface positions, surface depths, surface normals, surface albedos, and distances to the camera. For each pixel in a received image, the denoising application performs a first-order regression in a predefined neighborhood of the pixel to find a linear combination of pixel features that fits pixel colors in the predefined neighborhood. In such a first-order regression, the local regression weight of each pixel in the neighborhood may be determined using a metric which computes distances based on color values in patches around pixels being compared. In another embodiment, collaborative filtering may be performed in which filtered output from the first-order regression in each neighborhood is averaged with filtered output from overlapping neighborhoods to obtain a final output.

The invention relates to the field of testing technologies and instruments and particularly relates to a quartz flexible accelerometer temperature error calibration compensation method. Through carrying out eight-position calibration on a quartz flexible accelerometer at a room temperature and carrying out multiple groups of temperature-changing experiments on the quartz flexible accelerometer ina temperature box, output and temperature data of the accelerometer are obtained; an accelerometer temperature model method is then adopted to obtain a function relationship between the output of thequartz flexible accelerometer and the temperature; a back-stepping method and item-by-item differential processing are then adopted to obtain n-1 groups of item-by-item differential data results; a robust regression estimation method is finally adopted to obtain a function of a scale factor with respect to temperature; a function of zero bias with respect to temperature is then obtained; and a quartz flexible accelerometer output temperature error calibration compensation model is finally obtained. The cost can be effectively saved, and through a differential robust regression idea, a mutationpoint can be eliminated effectively and the high compensation accuracy can be obtained.

The invention relates to the field of medical detection and particularly discloses a noninvasive prenatal biological information detection and analysis method. For improving the accuracy of analyzing different quantities of to-be-detected samples, different detection and analysis methods are selected according to the different quantities of the to-be-detected samples, and different analysis policies are adopted for parameters obtained by the to-be-detected samples and parameters obtained by a normal reference set, so that the accuracy of analysis is improved to a greater extent. According to the method, the problem of inaccurate regression result caused by great influence of abnormal data on slope due to use of a least square method for regression in a process of correction by using a whole chromosome method in the prior art is well solved by adopting robust regression and CV regression, so that the robustness and accuracy of sample analysis are ensured. A set of analysis method for judging anomaly of sex chromosome by utilizing a ZZ value is originated; and the chromosome anomaly is judged by using the ZZ value method, so that related statistic judgment standards are better met, a result is more accurate, and the reliability of the method for judging the anomaly of the sex chromosome is enhanced.

The invention provides a method for predicting B[a]P (benzopyrene) in smoke of flue-cured tobacco strips based on Robust regression modeling. According to the method, a model from a physicochemical index item to smoke B[a]P is built through the existing flue-cured tobacco strip physicochemical data and the smoke B[a]P data; and for unknown flue-cured tobacco strip smoke B[a]P samples, physicochemical ingredient data of the flue-cured tobacco strip smoke B[a]P samples can be used for directly predicting the flue-cured tobacco strip smoke B[a]P value. The method provided by the invention has the advantages that the steps of rolling, burning, smoke catching, detection and the like in a traditional chemical mode are omitted; meanwhile, a Robust regression model is adopted; the defects caused by singular value samples in the physicochemical data or smoke data can be effectively avoided; and the robustness of the model is ensured to a great degree, and the point is the advantage of the Robust regression modeling superior to the ordinary linear regression modeling. Practice proves that the model can be used for effectively predicting the smoke B[a]P value, the detection efficiency is greatly improved, and the detection cost is reduced.

The invention discloses a robust-regression-based distributed photovoltaic generating electricity-stealing identification method. The method comprises the following steps: (1), establishing a historical information database; (2), carrying out determination and filtering on abnormal data existing in historical data; (3), carrying out processing by using a robust regression model algorithm to obtain an irradiation power curve; (4), carrying out operation to obtain a corresponding photovoltaic generation power; (5), carrying out calculation to obtain a theoretic generating capacity; and (6), carrying out determination. The method has the following beneficial effects: (1), with the robust regression model algorithm, the influence on the model precision by the abnormal data can be reduced and concrete modeling of a photovoltaic system inversion model and a photovoltaic conversion model can be avoided; and (2), electricity-stealing suspicion determination is carried out based on three-layer screening architecture, so that accuracy of abnormal determination of the photovoltaic electric quantity and the electricity-stealing determination reliability are high and the electricity-stealing checking pertinency is improved.

An analysis technique for genetic data to detect alternative spliced exons. Exon expression of similar data is analyzed using a robust regression technique to find outliers to the main regression. False outliers are detected and removed. The remaining outliers are identified as potential alternative splicing events.

The invention aims at solving various simulation conditions of background harmonic voltage fluctuation. Analyzing and comparing the advantages and disadvantages of the method and the traditional linear regression method, the invention provides a harmonic source responsibility division method based on CAE data screening, and the method comprises the steps: firstly dividing collected actual measurement data into a plurality of sections, carrying out the cross-approximation entropy calculation of the actual measurement harmonic voltage and current data of each section, and reserving the sectionsmeeting the requirements of a CAE threshold, so as to achieve the purpose of eliminating background harmonic interference; and then performing regression calculation on the reserved data by utilizingan M estimation robust regression method to avoid the influence of abnormal values on regression calculation to the greatest extent and obtain the harmonic impedance of the system, thereby realizing accurate harmonic responsibility division.

The invention discloses an early warning threshold setting method for dam safety monitoring data exception identification, and relates to the field of dam safety monitoring, and the method comprises the following steps: constructing a robust regression model based on a graph-based double-weight estimation function; on the basis of the robust regression model, replacing the residual standard deviation S with a scale estimator ST based on a position M estimator; constructing a predicted value confidence interval radius D based on a robust regression model, wherein the abnormal early warning threshold value is set to be [-3ST-D, 3ST + D]. According to the method, the accuracy of data exception online identification is improved, and the misjudgment and missed judgment rate of data exception online identification is reduced.

The invention provides a method for forecasting phenol in flue-cured tobacco smoke based on robust regression modeling. A model of physical and chemical index items and the smoke phenol is constructed according to existing flue-cured tobacco physical and chemical data and smoke phenol data, and a smoke phenol value of the flue-cured tobacco can be directly forecasted according to physical and chemical component data of an unknown flue-cured tobacco smoke phenol sample. According to the method, the steps of reeling, combustion, smoke catching, detection and the like of a conventional chemical method are eliminated; meanwhile, by the adoption of a robust regression model, the shortcomings caused by singular-value samples in the physical and chemical data or the smoke data can be effectively avoided; the robustness of the model can be guaranteed to an extremely large extent, and the robustness of the robust regression modeling is higher than that of common linear regression modeling. Practice shows that the smoke phenol value of the flue-cured tobacco can be effectively forecasted, the detection efficiency is greatly improved, and the detection cost is lowered.

The invention discloses a model-free tracking control method for piezoelectric ceramic actuators and a medium, relates to a robust regression prediction, PI control and self-adaptive inverse-based combined tracking control method, and belongs to the technical field of follow-up control. According to the method, the sine and cosine functions of main orders in micro vibration are taken as primary functions, robust regression is utilized to regress a reference control signal, and a regression value is utilized to predict a practical control signal so as to realize advanced control; the inner ringadopts PI control so as to realize preliminary smooth tracking, for the advanced control signal, of a piezoelectric ceramic; through calculating cross-correlation functions of the response of the piezoelectric ceramic and the reference control signal, an advanced quantity, for the reference control signal, of the response is estimated to adjust the delayed beat number of the pure lag links in a controller, thereby realizing the synchronization of the response of the piezoelectric ceramic and the reference control signal; and through calculating a linear relationship, for the reference controlsignal, of the response, the gain and zero offset of the controller are corrected by utilizing a correction coefficient, so that the amplitude tracking, for the reference control signal, of the response is realized.

The invention provides a method for predicting cured piece smoke NNK on the basis of robust regression modeling. A model from a physicochemical index item to the smoke NNK is built according to existing cured piece physicochemical data and smoke NNK data, and a cured piece smoke NNK value of an unknown cured piece smoke NNK sample can be directly predicted through physicochemical component data. By means of the method, the steps of rolling, burning, smoke capturing, detection and the like of a traditional chemical mode are omitted. Meanwhile, a robust regression model is adopted so that defects caused by singular values in the physicochemical data or in the smoke data can be effectively avoided, robustness of the model is guaranteed to a large extent, and compared with a common linear regression modeling, robust regression modeling has the advantage of better guaranteeing robustness. As is proved in practice, the cured piece smoke NNK value can be effectively predicted through the model, detection efficiency is greatly improved, and detection cost is lowered.

The invention provides a stable regression method for a remote sensing individual tree canopy and a forest diameter, and belongs to the technical field of the computer program and measurement. The method comprises the following steps: obtaining relevant data of the canopy and the forest diameter; establishing a stable regression model; carrying out an iterative determination on a stable regression model parameter of the individual tree canopy and the forest diameter. In a typical north east mid-temperature zone coniferous and broad-leaved mixed forest, based on extracting an individual tree canopy image of the high spatial resolution, by utilizing the Huber, an M- estimation method is proposed for solving the 'defect' that the traditional least square method is over sensitive for the data of the 'abnormal point', and the stable regression model of a remote sensing image individual tree canopy and the forest diameter is successfully established. The weighting factor distribution of an objective function is introduced through the M- estimation method, so that the 'defect' of the equivalent weighting of various samples (including the abnormal point) in the traditional least square method is changed. A menu program module of the stable regression model has the independence, the completeness, and the good portability.

The invention provides a robust regression modeling based dried slice flue gas ammonia prediction method. A model from physical and chemical index items to the flue gas ammonia is established through the existing dried slice physical and chemical data and flue gas ammonia data and a flue gas ammonia value of a dried slice can be directly predicted through the physical and chemical composition data of an unknown dried slice flue gas ammonia sample. According to the robust regression modeling based dried slice flue gas ammonia prediction method, the steps of rolling, combustion, flue gas capture, detection and the like which are performed by a traditional chemical method are omitted, meanwhile the robust regression model is adopted, the defects caused by singular value samples in the physical and chemical data or the flue gas data can be effectively overcome, the robustness of the model is ensured to a great extent, and accordingly the robust regression modeling is excellent in comparison with the ordinary linear regression modeling. Practice shows that the flue gas ammonia value of the dried slice can be effectively predicted, the detection efficiency is greatly improved, and the detection cost is reduced.

According to some embodiments, system and methods are provided, comprising building a first model structure for a reference domain; generating a first learned model for the first model structure using one or more data points associated with the reference domain; executing the first learned model with one or more data points in a target domain to predict a dependent variable associated with the target domain; calculating a residual variable for the predicted dependent variable associated with the target domain; building a second model structure for the target domain using the residual variable as a dependent variable; generating a second learned model for the second model structure using one or more data points associated with the target domain; and constructing a target model for the target domain, wherein the target model is the sum of the first and the second learned models. Numerous other aspects are provided.

The invention provides a method for predicting baked piece smoke hydrogen cyanide based on robust regression modeling. A model from the physical and chemical indicator items to smoke HCN is established through existing baked piece physical and chemical data and smoke HCN data, and for unknown baked piece smoke HCN samples, the baked piece smoke HCN value can be directly predicted through physical and chemical component data of the samples. By means of the method, the steps of coiling, combustion, smoke capture, detection and the like in a traditional chemical method are omitted; meanwhile, due to the adoption of a robust regression model, the defects caused by singular value samples in the physical and chemical data or the smoke data can be effectively overcome; compared with ordinary linear regression modeling, robust regression modeling has the superior advantage that robustness of the model is ensured to a great extent. The practice proves that the smoke HCN value in the baked pieces can be effectively predicted through the model, detection efficiency is greatly improved, and detection cost is reduced.

The invention provides a method for predicting the flue gas CO content of flue-cured tobacco slices based on robust regression modeling. A model from physicochemical indexes to flue gas CO is created according to existing physicochemical data and flue gas CO data of the flue-cured tobacco slices, and for an unknown flue gas CO sample of flue-cured tobacco slices, the flue gas CO content of the flue-cured tobacco slices can be directly predicted by physicochemical component data of the flue-cured tobacco slices. According to the method, the steps of winding, burning, gas capture, detection and the like in a conventional chemical mode are removed; meanwhile, a robust regression model is adopted, so that the disadvantages caused by singular value samples in the physicochemical data or the flue gas data can be effectively avoided, the robustness of the model is guaranteed to a very great extent, and the robust regression modeling is superior to general linear regression modeling. The practice proves that the model can effectively predict the flue gas CO content of the flue-cured tobacco slices, the detection efficiency is greatly improved, and the detection cost is reduced.

The invention provides an image segmentation method adopting a regression algorithm. The main contents of the method comprise a background fitting smoothing model, pixel regression, RANSAC robust regression, and an integral segmentation algorithm. The difficulties of the prior art are overcome by adopting a regression segmentation algorithm; firstly a background part is subjected to smoothing processing through the fitting smoothing model; secondly the pixel intensity is predicted through the pixel regression, and optimal values of model parameters are found, so that the background is not influenced by pixels of a foreground; thirdly the number of abnormal values is reduced to the maximum extent through an iterative method by the RANSAC robust regression technology, and the algorithm speed is increased through preprocessing; and finally all the pixels are fitted by adopting a least square method through the integral segmentation algorithm, so that the image segmentation is finished. The algorithm provides excellent image segmentation performance, the difficulties of common methods before are overcome, and a segmentation result also has relatively good performance when background and foreground image ranges are overlapped.

The disclosure provides an approach for denoising (also referred to as “filtering”) rendered images. In one embodiment, a denoising application takes as input rendered images and feature buffers that encode image information such as surface positions, surface depths, surface normals, surface albedos, and distances to the camera. For each pixel in a received image, the denoising application performs a first-order regression in a predefined neighborhood of the pixel to find a linear combination of pixel features that fits pixel colors in the predefined neighborhood. In such a first-order regression, the local regression weight of each pixel in the neighborhood may be determined using a metric which computes distances based on color values in patches around pixels being compared. In another embodiment, collaborative filtering may be performed in which filtered output from the first-order regression in each neighborhood is averaged with filtered output from overlapping neighborhoods to obtain a final output.

The invention provides a robust regression modeling based dried slice flue gas ammonia prediction method. A model from physical and chemical index items to the flue gas ammonia is established through the existing dried slice physical and chemical data and flue gas ammonia data and a flue gas ammonia value of a dried slice can be directly predicted through the physical and chemical composition data of an unknown dried slice flue gas ammonia sample. According to the robust regression modeling based dried slice flue gas ammonia prediction method, the steps of rolling, combustion, flue gas capture, detection and the like which are performed by a traditional chemical method are omitted, meanwhile the robust regression model is adopted, the defects caused by singular value samples in the physical and chemical data or the flue gas data can be effectively overcome, the robustness of the model is ensured to a great extent, and accordingly the robust regression modeling is excellent in comparison with the ordinary linear regression modeling. Practice shows that the flue gas ammonia value of the dried slice can be effectively predicted, the detection efficiency is greatly improved, and the detection cost is reduced.

One embodiment of the invention discloses a track prediction method based on robust regression. The method comprises the following steps: S10, obtaining a measurement value of an information source; S13, calculating a prediction coefficient; S15, calculating a weighted prediction coefficient; and S17, judging whether the number of iterations is met or not, if so, ending the process, and if not, carrying out step S15. According to the method, different weights are given to the measurement data according to the measurement residual errors, so that the influence of outliers on prediction is reduced, and the prediction precision is improved.