30 results about "Residual correction" patented technology

The present invention provides a disk fault processing method in a disk array system and a data reconstruction method, including: the disk of the disk array generates read-write faults; the fault data block is located, the residual correction data of the fault disk is retained, the fault information table is checked, the strip of the faulty data block whether existing other fault data blocks or not is judged, if yes, the disk array system is provided to a ''read-only mode'', forbidding user execute write operation, if not, the position information of the fault data block is recorded in the fault information table, then whether a redundancy data disk existing in the disk array system or not is judged, if not, entering in the ''degradation'' operation mode, otherwise, the fault data block is processed with data reconstruction, recovering the fault data.

The invention relates to a urban electricity consumption prediction method based on a GM(1,1) model and a grey Verhulst model. The method includes the following steps: firstly performing cos(x) conversion on historical electricity consumption data to enhance the smoothness of a data sequence so as to increase the prediction precision of the GM(1,1); then inputting the data which is obtained by the cos(x) conversion to the GM(1,1) model, obtaining a predicted electricity consumption data sequence; and eventually, inspecting the precision of the electricity consumption data sequence, if the a prediction value does not exceed the precision requirement, outputting the electricity consumption data sequence; if the prediction value exceeds the precision requirement, in combination with the characteristics of the changes of a residual sequence and the advantages of the grey Verhulst model, establishing a grey Verhulst model, performing residual correction on the residual sequence of low or lower precision, and until the prediction value of the electricity consumption meets the precision requirement, outputting the corrected electricity consumption data sequence. Compared with prior art, according to the invention, the method is advantaged by being capable of better predicting urban annual electricity consumption and higher prediction precision.

A method for improving color calls or base calls utilizes current and prior cycle multi-channel intensity data from a sequencing run to model residual cycle buildup. The model is applied to correct the multi-cycle channel intensity for the current cycle. The corrected multi-cycle channel intensity is used for color calls or base calls for the current cycle.

The invention provides a GNSS occultation ionospheric residual correction method and system, a device and a storage medium. The method comprises the steps of: preprocessing GNSS occultation original observation data, vTEC maps data and geomagnetic field observation data to obtain GNSS occultation geometric data, ionospheric data and geomagnetic field data; based on the GNSS occultation geometric data, a three-dimensional NeUoG ionospheric mode and the ionospheric data, calculating an electron density profile at the location of an ionospheric puncture point on an incident ray side and an emergent ray side; based on the GNSS occultation geometric data, an IGRF geomagnetic field mode and the geomagnetic field data, calculating a geomagnetic field intensity profile along a signal path at the location of the ionospheric puncture point on the incident ray side and the emergent ray side; based on the electron density and the geomagnetic field intensity along the signal path, calculating a bending angle ionospheric residual profile. The method of the invention can be used in the atmospheric parameter inversion of a single GNSS occultation event, weakens the influence of the ionospheric residual, obtains a high-precision GNSS occultation curved angular profile, and is efficient and reliable.

The invention provides a three-dimensional high-precision bin fractionation processing and evaluation technology for seismic data, and the method comprises the following steps: analyzing original data, and carrying out trace header modification, grid definition and linear normal moveout correction on data acquired by an observation system, thus acquiring a prestack multi-domain composite denoised image; based on the relationship between bin fractionation and the signal-to-noise ratio as well as the relationship between bin fractionation and the multiplicity during forward modeling, comprehensively compensating for the vibration amplitude, thus acquiring a tomographic inversion refracted wave static correction image; analyzing the quantitative relationship between bin fractionation and the longitudinal resolution and lateral resolution, carrying out prestack shot-domain optimized deconvolution, and carrying out azimuth restrained bin fractionation and speed optimized analysis, thus acquiring a progressive frequency-divided residual correction image; and carrying out multiple three-dimensional prestack migrations, and establishing a three-dimensional prestack migration speed model, thus acquiring a prestack migration image. Thus, the method provided by the invention provides a reliable guarantee for improving the exploration development precision and upgrading a seismic exploration technology.

The invention discloses an image pixel marking method based on deep convolution neural network. The main content of the method comprises the following steps of image input, detection, replacement, refinement and prediction mark estimation. The process comprises the following steps that a traffic scene image and initial estimations for the image are inputted by using the deep convolution neural network, detection components detect errors in the input marks, and then substitute components replace the error marks with new marks, at last, all output marks are improved overall in the manner of residual correction, and new, accurate mark estimations are acquired. The image pixel marking method can save a lot of memory and time by adopting a neural network model and achieve more accurate results considering the dependency which exists in a joint space of input and output variables, objective functions are applied in the final outputs, and the end-to-end learning of a dense image marking task is allowed.

An image stabilization apparatus including a supporting unit that supports an image capturing unit, a rotation unit that rotates the supporting unit, a calculation unit that calculates a residual correction amount of image blur after image blur correction is performed by a first image blur correction unit, a second image blur correction unit that corrects the residual correction amount of image blur using a method different from that of the first image blur correction unit, and a change unit that changes allocation between a correctable range of the second image blur correction unit correcting image blur in a horizontal or vertical direction on the imaging plane and a correctable range of the second image blur correction unit, based on a rotation angle of the rotation unit and the residual correction amount of image blur.

The invention provides a gas in oil prediction method via combined prediction and residual correction. The method comprises the following steps: (1) separating a gas through an oil-gas separation device, analyzing gas composition and proportion by using dissolved gas online chromatographic analysis and gas sensitive sensor analysis, and establishing a dissolved gas time sequence; (2) determining the correlation between gases by using a gray correlation degree, establishing an MGM (1, n) prediction model, and outputting a gray prediction result; and (3) determining a training set and a test set, and initially correcting the training set and test set results through residual adaptive regression, performing state division on the training set on the basis, and further correcting a predicted value of the test set by using a Markov model. The gas in oil prediction method provided by the invention is mainly applied to the state monitoring of an oil-filled transformer so as to discover a faultin time and to avoid further loss conveniently, and the error caused by the corrected predicted value is obviously smaller than the error caused by the traditional gray prediction result;and accurateand efficient fault identification is ensured, the principle is simple, and the operation is simple and convenient.

The invention relates to a scene segmentation correction method. A global residual correction network is adopted, a completely residual convolutional network is used as a front-end model, a confidencemap and an original image of the front-end model are spliced according to a channel as an input of the global residual correction network so as to output a global correction residual, and the globalcorrection residual is added to the confidence map to obtain a correction result of scene segmentation; and the global residual correction network is trained by using a known scene segmentation data set. Meanwhile, the method further proposes serially connecting the global residual correction network and a local boundary correction network to form a cascade frame, and the cascade frame can be usedfor global correction and local correction on the segmentation result of the front-end model, thus obtaining a more accurate scene segmentation correction result.

The invention discloses a residual correction method for NWP inversion tropospheric delay under multi-factor constraints. The method comprises steps: continuous one-year ZTD of all continuous runningreference stations in a working area and continuous one-year re-analysis data provided by the NWP in the area are firstly acquired; the ZTD of all continuous running reference stations in the area areinverted by using an integral method; the NWP inversion ZTD residual is fit based on a polynomial fitting method based on the minimum absolute residual method, and NWP prediction data and the initialcoordinates of an observation station at the current moment are acquired; by using the above residual model, the NWP_ZTD residual after correction at the current moment is calculated; the initial value of the NWP_ZTD at the current moment is inverted by using the NWP prediction data at the current moment; the initial value of the above NWP_ZTD and the NWP_ZTD residual after correction are added,and the accurate NWP_ZTD at the current moment is obtained. The method performs modeling for the first time based on the relationship between temperature, humidity, position and other factors and a tropospheric residual time sequence, the NWP inversion tropospheric delay residual is improved to improve the tropospheric delay estimation precision to a sub-centimeter level.

The invention discloses a residual correction method for wind speed prediction based on a GARCH (Generalized ARCH) model, belongs to the technical field of power transmission and distribution, and aims to improve an original wind speed prediction method and improve the wind speed prediction precision. The technical scheme is as follows: adopting the GARCH model to predict an original wind speed series; establishing a regression model for a residual fitted by the GARCH model, and using the regression model to predict the following residuals; and finally, using a predicted value to correct a preliminary prediction result of the GARCH model. The experimental result shows that the residual correction method for the wind speed prediction based on the GARCH model provided by the invention is better than a prediction method of a traditional ARMA and a prediction method of a BP neural network model, and has higher prediction accuracy.

The invention relates to a scene segmentation correction method. A local boundary correction network is adopted, a completely residual convolutional network is used as a front-end model, a confidencemap and an original image of the front-end model are spliced according to a channel as an input of the local boundary correction network so as to output local polymerization coefficients of all positions of the confidence map, the local polymerization coefficients are multiplied with the corresponding positions of the confidence map, and the products are polymerized to a center point to obtain a local boundary correction result of scene segmentation; and the local boundary correction network is trained by using a known scene segmentation data set. Meanwhile, the method further proposes serially connecting a global residual correction network and the local boundary correction network to form a cascade frame, and the cascade frame can be used for global correction and local correction on thesegmentation result of the front-end model, thus obtaining a more accurate scene segmentation correction result.

The invention discloses a long-term power load prediction method based on a residual correction grey prediction model, which is applied to the technical field of power system load prediction, and aimsto solve the problems of enhanced volatility along with load change and poor fitting and prediction effects of an existing grey prediction model. Under the condition that the power load sequence of the previous n years is acquired, the power load value of the (n + 1) th year is predicted by using a linear time-varying parameter discrete grey TDGM (1, 1) model, then an original prediction model iscorrected by using a Fourier series residual correction method, and finally, a corrected analog value and a corrected predicted value are obtained; the corrected TDGM (1, 1) model has higher fittingand prediction precision, and the adaptability and flexibility of the grey prediction model are improved.

An image stabilization apparatus including a supporting unit that supports an image capturing unit, a rotation unit that rotates the supporting unit, a calculation unit that calculates a residual correction amount of image blur after image blur correction is performed by a first image blur correction unit, a second image blur correction unit that corrects the residual correction amount of image blur using a method different from that of the first image blur correction unit, and a change unit that changes allocation between a correctable range of the second image blur correction unit correcting image blur in a horizontal or vertical direction on the imaging plane and a correctable range of the second image blur correction unit, based on a rotation angle of the rotation unit and the residual correction amount of image blur.

The invention provides a track quality prediction method and system based on an improved gray combination model, and the method comprises the steps: preprocessing original data, and also comprises thefollowing steps: constructing an improved non-equal-time-interval gray IGM prediction model; optimizing the weight and the threshold of the Elman network by using a PSO algorithm, and constructing aPSO-Elman neural network correction model; and constructing an IGM-PSO-Elman model by integrating the advantages of the improved non-equitime-interval grey IGM model, the particle swarm optimization PSO algorithm and the Elman neural network. The invention provides the orbit quality prediction method and system based on the improved grey combination model; improvement is carried out on the basis of a traditional gray model, a weight coefficient optimization first-order accumulation generation process is introduced, weight matrix optimization model parameter solving is added, calculation of a difference equation background value is optimized by utilizing an integral area, and a PSO-Elman residual error correction model can reduce data fluctuation and reduce an improved non-equal-time-interval gray prediction residual error.

The invention discloses a cosine function and power function combined transformation-based grey prediction model method for initial sequences. The method comprises the following steps of: 1, generating an original data sequence; 2, preprocessing an initial sequence; 3, carrying out cosine function and power function combined transformation on the preprocessed sequence; 4, generating a cumulative sequence; 5, carrying out grey prediction; 6, determining a response function; 7, restoring a prediction value; 8, calculating a relative error; 9, selecting a minimum relative error value; and 10, carrying out residual correction on a minimum relative error value sequence to obtain a corrected sequence. The method is capable of being effectively adapted to various data change types and effectivelyimproving prediction precision of models.

The invention discloses an unsupervised hyperspectral image blind fusion method and system based on a spatial-spectral joint residual correction network. The method comprises the following steps: building a degradation network structure of a hyperspectral image, in order to simulate a space and spectrum downsampling process; establishing a space and spectrum residual fusion network model, and using, a difference value between a low-resolution result obtained through the degradation model and training data, as input of a fusion network, namely fusing residual errors in space and spectrum dimensions to obtain a residual plot corresponding to the input data; correcting initialized data, and sending a corrected result into the degradation network and a spatial-spectral joint correction network for multiple iterations to improve the precision of a fusion result. According to the invention, a spatial-spectral joint correction network suitable for unsupervised hyperspectral image blind fusion is used, and the spatial-spectral joint correction network can obtain an error graph between an input hyperspectral image and a true value.

The present invention provides a high-precision time-frequency signal dynamic control method based on residual correction, the residual serves as the correction amount to correct the time difference model and the adjustment amount, the time difference prediction precision is improved, under the active condition, the synchronization precision with a reference source is effectively improved, and under the passive condition, through adoption of the dynamic control method, phase deviation caused by signal frequency deviation and aging is avoided to the maximum extent, so that the phase of the signal source changes slowly and stably, signal self-maintenance is further realized, high-precision synchronization of the signal source and the reference source is ensured, and the problems that the performance of the system is reduced and services and tests cannot be provided normally due to abnormality of the reference source are solved.

The invention relates to a weak light image enhancement method and device based on a conditional adversarial network, computer equipment and a storage medium. The method comprises the following steps: using a plurality of convolutional neural networks of a residual structure connected in series as a generator and a WGAN-GP as a discriminator, performing down-sampling on a weak light image through a stepping convolutional network, processing a down-sampled feature map through a residual network, performing up-sampling on a residual correction feature map through a transposition network, fusing the weak light image sample original image and the up-sampling feature image through a jump connection structure and then activating and outputting, and obtaining a generator generation image; judging the generator generation graph through a discriminator network to obtain a discriminator output matrix, and generating an effective weak light enhancement model under the constraint action of a preset loss function. According to the model, detail information of an original image can be reserved, the brightness of the image is improved, meanwhile, gain noise interference cannot be generated, and the generated image is more natural and high in quality.

A method for predicting sinter tumbler strength comprises the steps of building a prediction model and using the built prediction model to predict the tumbler strength. The step of building the prediction model comprises a step of collecting sampling data of various chemical components with the sinter tumbler strength, a step of building a gray residual correction model and a support vector machine model to predict the tumbler strength aiming at the sampling data, and a step of carrying out combined prediction: confirming optimal weight coefficients of the gray residual correction model and the support vector machine model to form an optimal combined prediction model to predict the tumbler strength. By means of the method, the combined prediction of the gray residual correction model and the support vector machine model is finally achieved, prediction accuracy is high, and robustness and generalization is better.

The invention discloses a correction liquid bottle, belongs to a container, and particularly relates to a bottle containing correction liquid for users to correct characters. The correction liquid bottle comprises a bottle body and a bottle cap, and is characterized in that a conical nozzle is arranged at the top of the bottle body, an end face of a liquid outlet at the top of the nozzle is rectangular, the bottle cap is in threaded connection with the bottle body, and a conical cleaning pad is mounted on the inner top of the bottle cap. The correction liquid bottle has the advantages that since the end face of the liquid outlet at the top of the bottle nozzle is rectangular, correction liquid squeezed from the liquid outlet is in the shape of strip with equal width and is uniform in thickness, and portions corrected by the correction liquid are tidy and beautiful; by the conical cleaning pad mounted on the inner top of the bottle cap of the correction liquid bottle, residual correction liquid at the mouth of the correction liquid bottle can be wiped and cleaned quickly, and users can use the correction liquid bottle the next time conveniently. Technical modification can be realized for existing enterprises conveniently due to the fact that only the bottle nozzle and the bottle cap of a conventional correction liquid bottle are required to be modified.

In a method for correcting for residual activity due to an earlier tracer in a later PET or SPECT scan image at reconstruction, thereby generating a residual-corrected later image the residual activity is estimated by detecting the time of an introduction of a tracer for the later PET or SPECT scan; and separating the residual activity from the true counts during iterative reconstruction of the PET or SPECT scan image.

The invention discloses a gray system submarine oil and gas pipeline drop object corrosion reliability evaluation method, comprising: establishing a gray differential equation, calculating gray system parameter values ​​according to pipeline corrosion depth by least square estimation, and using a traditional gray system to perform a simulation Calculate the simulated value; then, according to the inconsistency between the simulated derivative recovery value and the cumulative reduction value, in order to reduce the error caused by the reciprocating operation, the residual error is used to correct the simulated value for a secondary simulation, and the primary predicted value and the secondary predicted value are compared. Values ​​are combined to obtain a complete subsea pipeline corrosion correction model for prediction, thereby improving the prediction accuracy of pipeline corrosion depth. Finally, taking a certain subsea pipeline test section as an example, using the gray system of tail section residual error correction to predict the pipeline corrosion depth Through simulation, compared with the traditional gray system, the prediction accuracy has been significantly improved, so that the pipeline can be repaired or replaced in time, thereby reducing the resulting economic losses and safety hazards.

A method for improving color calls or base calls utilizes current and prior cycle multi-channel intensity data from a sequencing run to model residual cycle buildup. The model is applied to correct the multi-cycle channel intensity for the current cycle. The corrected multi-cycle channel intensity is used for color calls or base calls for the current cycle.

The present invention provides a disk fault processing method in a disk array system and a data reconstruction method, including: the disk of the disk array generates read-write faults; the fault data block is located, the residual correction data of the fault disk is retained, the fault information table is checked, the strip of the faulty data block whether existing other fault data blocks or not is judged, if yes, the disk array system is provided to a ''read-only mode'', forbidding user execute write operation, if not, the position information of the fault data block is recorded in the fault information table, then whether a redundancy data disk existing in the disk array system or not is judged, if not, entering in the ''degradation'' operation mode, otherwise, the fault data block is processed with data reconstruction, recovering the fault data.

The invention discloses a residual correction method for NWP inversion tropospheric delay under multi-factor constraints. The method comprises steps: continuous one-year ZTD of all continuous runningreference stations in a working area and continuous one-year re-analysis data provided by the NWP in the area are firstly acquired; the ZTD of all continuous running reference stations in the area areinverted by using an integral method; the NWP inversion ZTD residual is fit based on a polynomial fitting method based on the minimum absolute residual method, and NWP prediction data and the initialcoordinates of an observation station at the current moment are acquired; by using the above residual model, the NWP_ZTD residual after correction at the current moment is calculated; the initial value of the NWP_ZTD at the current moment is inverted by using the NWP prediction data at the current moment; the initial value of the above NWP_ZTD and the NWP_ZTD residual after correction are added,and the accurate NWP_ZTD at the current moment is obtained. The method performs modeling for the first time based on the relationship between temperature, humidity, position and other factors and a tropospheric residual time sequence, the NWP inversion tropospheric delay residual is improved to improve the tropospheric delay estimation precision to a sub-centimeter level.

The invention belongs to the technical field of heat treatment, and particularly relates to a method for reducing heat treatment deformation of sheet metal angle parts. After being subjected to heat treatment, the sheet metal angle parts tend to generate large deformation (such as warping and distortion) and are large in elasticity and difficult in correction; and the surface of the parts is freeof machining allowance, the residual correction marks on the surface of the parts cannot be removed, and the appearance quality of the parts is influenced. According to the method for reducing heat treatment deformation of the sheet metal angle parts, the same preformed parts are arranged on the two sides of a reinforcing plate in a mode of symmetry welding / tack welding, and welding parts / tack welding parts are subjected to heat treatment, and then are separated and machined to obtain final parts; deformation amount of the parts after heat treatment is greatly reduced, the correcting workloadis reduced, and the production efficiency is improved, the rejection rate of the parts is reduced, and meanwhile, the surface quality of the parts is improved, difficulty and workload of correction ofthe parts are reduced, rejection of the parts is avoided, and material consumption is reduced.

The invention discloses a power load prediction method of an improved gray model based on particle swarm optimization, belongs to the technical field of power system load prediction, and aims at solving the problems that an existing gray prediction model is large in fluctuation along with load change and poor in prediction effect, the method comprises steps of utilizing agray prediction model DTGM (1, 1) based on data transformation, predicting a load value of an (n + 1) year under the condition of obtaining load data of previous n years, solving optimal values of residual state interval parameters lambda1, lambda2,... lambdan through a particle swarm optimization (PSO) algorithm, obtaining an optimal residual correction value to correct an original prediction model, and finally obtaining a corrected prediction value; the corrected DTGM (1, 1) has higher prediction precision, and the defects of a traditional gray prediction model are overcome.

Decoder for decoding a picture from a data stream, configured to check whether a plurality of coding parameters, which are contained in the data stream, relate to a predetermined portion of the picture and control a prediction residual transform mode and a quantization accuracy with respect to the predetermined portion, are indicative of a coding parameter setting corresponding to a lossless prediction residual coding. Responsive to the plurality of coding parameters being indicative of the coding parameter setting corresponding to the lossless prediction residual coding, the decoder is configured to set one or more predetermined coding options relating to one or more tools of the decoder for processing a prediction residual corrected predictive reconstruction with respect to the predetermined portion so that the one or more tools are disabled with respect to the predetermined portion.