112 results about "Steep descent" patented technology

A novel apparatus for and a method of estimating, calibrating and tracking in real-time the gain of a radio frequency (RF) digitally controlled oscillator (DCO) in an all-digital phase locked loop (ADPLL). Precise setting of the inverse DCO gain in the ADPLL modulating path allows direct wideband frequency modulation that is independent of the ADPLL loop bandwidth. The gain calibration technique is based on a steepest descent iterative algorithm wherein the phase ADPLL error is sampled and correlated with the modulating data to generate a gradient. The gradient is then scaled and added to the current value of the DCO gain multiplier.

A novel apparatus for and a method of estimating, calibrating and tracking in real-time the gain of a radio frequency (RF) digitally controlled oscillator (DCO) in an all-digital phase locked loop (ADPLL). Precise setting of the inverse DCO gain in the ADPLL modulating path allows direct wideband frequency modulation that is independent of the ADPLL loop bandwidth. The gain calibration technique is based on a steepest descent iterative algorithm wherein the phase ADPLL error is sampled and correlated with the modulating data to generate a gradient. The gradient is then scaled and added to the current value of the DCO gain multiplier.

A method of creating a super-resolved color image from multiple lower-resolution color images is provided by combining a data fidelity penalty term, a spatial luminance penalty term, a spatial chrominance penalty term, and an inter-color dependencies penalty term to create an overall cost function. The data fidelity penalty term is an L1 norm penalty term to enforce similarities between raw data and a high-resolution image estimate, the spatial luminance penalty term is to encourage sharp edges in a luminance component to the high-resolution image, the spatial chrominance penalty term is to encourage smoothness in a chrominance component of the high-resolution image, and the inter-color dependencies penalty term is to encourage homogeneity of an edge location and orientation in different color bands. A steepest descent optimization is applied to the overall cost function for minimization by applying a derivative to each color band while the other color bands constant.

The invention discloses a rapidly converged scene-based non-uniformity correction method, wherein the aim of non-uniformity correction is achieved by minimizing interframe registration error of two adjacent images. The method mainly comprises the following steps of: initializing gain and offset correction parameters and acquiring an uncorrected original image; acquiring a new uncorrected original image, and carrying out non-uniformity correction on the new uncorrected original image and the previous uncorrected original image by utilizing the current non-uniformity correction parameters; obtaining relative displacement, scene correlation coefficient and interframe registration error of two corrected images by utilizing an original point masking phase correlation method; and updating correction parameters along the negative gradient direction by adopting a steepest descent method. The method disclosed by the invention has the advantages of high correction accuracy, fast convergence speed, no ghost effect and low calculated amount and storage content and is especially applicable to being integrated into an infrared focal plane imaging system, and the effect of improving imaging quality, environmental suitability and time stability of an infrared focal plane array is achieved.

The invention relates to a method for identifying practical parameters of a synchronous generator on basis of a load rejection test and a numerical difference. The method comprises the following steps: utilizing a conventional method to confirm steady parameters of a generator; taking an exciting voltage after the load rejection test is performed as an input volume; taking a difference between a theoretical value of a machine-end voltage and a practically measured value under an evaluating system as a target function; and utilizing an iterative solution method for an extreme value of the target function based on a steepest descent method to realize the identification for the parameters of the generator, wherein a numerical difference method is used for obtaining the gradient of the target function during an iterative process and a difference value of adjacent twice target functions and the gradient of the target function are taken as control conditions for ending the iteration. The load rejection test, on which the identification is based, is easily realized; a high enough induction current can be generated in a damping winding after the load rejection test is performed; the machine-end voltage is obviously changed during the temporary state process and the sub-temporary state process; the interference of environmental noise to the identification for a time domain is efficiently overcome; and the problem of unsteady indentifying result is solved.

The invention discloses an image registration method based on a synthetic aperture radar (SAR) image and digital elevation model (DEM) data, wherein the problems that the positioning precision to ground target points in SAR image is not high and positioning is difficult to realize in actual projects in the prior art are mainly solved. The method comprises the following steps of: solving a longitude L, a latitude B and an elevation value h' of a ground target point in a spherical coordinate system by using a steepest descent method according to the Range-Doppler (RD) principle and the earth model equation, and transforming the values into a geodetic coordinate system; looking up the elevation value h' corresponding to B and L in a DEM database, and comparing the absolute value of h'- h' with a set threshold value epsilon, if the absolute value of h'- h' is greater than epsilon, enabling h to be equal to h' and returning to resolve the equation system again, and if the absolute value ofh'- h' is identical to or less than epsilon, stopping iteration, and taking h' as the elevation value of the ground target point, and the corresponding latitude B and longitude L as the latitude and longitude of the ground target point; linearly correcting the longitude and latitude to eliminate system errors, so that a final ground target point positioning result is obtained. The image registration method disclosed by the invention has the advantages of small operand and high positioning precision, and can be used for positioning ground target points in SAR images in actual projects.

A daily electricity consumption prediction method based on an artificial neural network belongs to the power supply technology field. A topology structure of a neural network model comprises an input layer, a hidden layer and an output layer, and the learning rules of the neural network model are that: a parallel distribution processing model of an error back propagation algorithm is adopted, a steepest descent method is used, and the weight and the threshold value of the network are adjusted continuously by the back propagation, so that the error sum of squares of the network is minimum. By the training sample learning, a prediction model can be established automatically and is hidden inside the artificial neural network, by using the electricity consumption prediction method in a short-term load predication, a time sequence model can be supplemented beneficially. The daily electricity consumption prediction method based on the artificial neural network can be widely used in the load predication and load management field of an electric power system.

The invention provides a light source-mask mixed optimizing method based on an Abbe vector imaging model. The method comprises the following steps of: setting a light source image pixel value and transmittances of a mask opening part and a light-stopping part, setting variable matrixes [omega]S and [omega]M, constructing a target function D into a square of an Euler distance between a target image and an image formed in the corresponding photoresist of the current light source and mask, and guiding the mixed optimization of the light source and mask image by using the variable matrixes [omega]S and [omega]M and the target function D. Compared with the traditional light source single optimizing method, mask single optimizing method, light source-mask simultaneous optimizing method and light source-mask alternative optimizing method and the like, the method disclosed by the invention can more effectively improve the resolution ratio of a photoetching system. Simultaneously, the light source and mask optimized by the method disclosed by the invention are applicable to conditions of small numerical aperture (NA), as well as applicable to the conditions that NA is greater than 0.6. Moreover, according to the invention, light source images and mask images are optimized through optimizing the gradient information of the target function and combining with a steepest descent method, and the optimizing efficiency is high.

The invention discloses a method for analyzing a running trend of an electric power communication transmission network. The method is characterized by comprising the following steps: (1) acquiring basic data, and providing dynamic parameters and static parameters for a trend analysis model; (2) performing mutation analysis on performance; judging whether the acquired dynamic parameters are out of limit or not based on a performance parameter early warning threshold value; directly generating a performance alarm if the acquired dynamic parameters are out of limit; (3) performing trend analysis when the performance is steadily changed; constructing an error rate and optical power analysis model according to dynamically acquired performance parameters by adopting secondary dynamic exponential smoothing; obtaining a trend analysis weight through a steepest descent iteration method by combining the dynamic parameters and historical data, and further, constructing a linear prediction equation; (4) performing early warning computation, and outputting a performance prediction result. The method disclosed by the invention is applied to a communication management system, and therefore, a powerful data support is provided for the whole-process intelligent computational analysis from data acquisition, trend prediction, early warning analysis and a maintenance strategy.

The invention discloses a route planning method for multiple unmanned aerial vehicle cooperative tracking under multiple constraints and can help overcome a problem that unmanned aerial vehicle physics and threat / obstacle constraints are not taken into consideration in conventional multiple unmanned aerial vehicle cooperative tracking. The method is characterized in that distributed fusion rules are used for fusing local filtering estimation of an object by all unmanned aerial vehicles, and an A optimal rule is used for establishing a cost function; as for the threat / obstacle constraints, thecost function is corrected via a penalty function, a steepest descent method is used for rapidly solving an optimization problem, and flight of the unmanned aerial vehicles can be facilitated via constraints over maximum turning angles of the unmanned aerial vehicles. The method can be help effectively solve a problem of multiple unmanned aerial vehicle cooperative tracking under thread / obstacle and physics constraints, and therefore cooperative tracking of the object performed by the multiple unmanned aerial vehicles in a complex environment can be realized.

The invention discloses an auto-fluorescence tomography re-establishing method based on a multiplier method. The auto-fluorescence tomography re-establishing method based on the multiplier method comprises the following steps of: carrying out discretization by using a finite element method diffusion equation, and establishing an optimization problem model without constraint conditions based on a penalty term of an L1 norm; obtaining a dual model of the optimization problem model without constraint conditions; establishing an augmentation lagrange function of the dual model; simplifying the maximum function of the augmentation lagrange function; solving the maximum value of the augmentation lagrange function by using a truncated-Newton algorithm; upgrading the target vector by using the gradient of the augmentation lagrange function as the steepest descent direction of a target vector; upgrading a penalty vector; and calculating an objective function value J(w), calculating k=k+1 if the ratio of the norm of J(w)k-J(w)(k-1) to the norm of Pai m being not smaller than t0l is real, and jumping to the step S4, otherwise, ending the calculation, wherein t0l is the convergence efficiency threshold value of the target function. The auto-fluorescence tomography re-establishing method provided by the invention can quickly obtain accurate and reliable light source distribution information within a large image-forming region, so that other parameters except from the regularization parameter can realize self-adaptive adjustment for improving the image-forming robustness.

The invention discloses an interference SAR data imaging and elevation estimation method. On the basis of the back-projection imaging theory, by adopting the corresponding relation between interferometric phase and terrain elevation in the interference SAR back-projection algorithm imaging process, and by using the principle that the minimum interferometric phase is obtained when the imaging space terrain elevation is the real terrain elevation, the terrain elevation corresponding to the minimum interferometric phase is found by the steepest descent method, and interference SAR data is projected to the optimal imaging space terrain elevation which is estimated and obtained by the steepest descent method, for the purpose of imaging. In this way, synchronous processing of interference SAR echo data imaging and elevation estimation is realized. Compared with the prior art, the invention reduces interference SAR image distortion effect which may be caused due to terrain elevation fluctuation, and improves the interference SAR imaging image quality and the terrain elevation estimation precision under the condition of fluctuating terrain elevation.

The invention relates to a permanent magnet synchronous motor parameter identification method based on an artificial neural network. The two-layer linear neural network is adopted for the method, and a motor vector control strategy is used in cooperation. First, the neural network is used for carrying out off-line training on acquired motor current rotating speed data, after the control accuracy meeting the requirement is achieved, a neural network weight obtained through off-line training serves as an initial value for on-line learning of the neural network, and then on-line learning adjustment is carried out on a system. The weight of the neural network is adjusted on line through a steepest descent method so that the output of the neural network can be close to an actual value, and then the stator resistance and the quadrature axis and direct axis inductance of a motor and the estimated value of rotor flux linkage are obtained.

The invention provides an air pollution tracing and trend prediction method with visual big data. The method comprises the following steps: using air diffusion conditions as input factors, and establishing a feedforward model for air pollution tracing and trend prediction by using a multilayer feedforward neural network trained by an error back propagation algorithm; and executing a learning process of the feedforward model, wherein historical air pollution data are used as the input training factors of the feedforward model, and gradually adjusting the weight and the threshold of the feedforward model of the network via the error back propagation by using the steepest descent method of an error performance function, so that the output errors of the feedforward model of the network are reduced continuously.

The invention provides a method for optimizing optical proximity correction (OPC) of a nonideal photoetching system based on an Abbe vector imaging model. The method comprises the following steps of: setting transmittivity of an opening part and a light blocking part in a mask; setting a variable matrix Omega; setting a target function D as linear combination of an imaging evaluation function of an ideal image surface and an imaging evaluation function of an image surface of which the defocusing quantity is fnm; and guiding optimization on a mask pattern by using the variable matrix Omega andthe target function D. By using the vector imaging model and taking vector characteristic of an electromagnetic field into consideration during acquisition of a space image, the optimized mask is suitable for the photoetching system with small numerical aperture (NA) and also suitable for the photoetching system of which the NA is more than 0.6. By the method, the gradient information of optimizing the target function is utilized and a steepest descent method is combined to optimize the pattern and the phase of an attenuated phase-shifting mask, so the optimization efficiency is high.

In accordance with a particular embodiment of the present invention, a method is offered that includes characterizing a data correlation matrix for an idle pattern offline in a filter environment and, further, using a static adaptive control scheme with a static value of a data-pattern compensation matrix to achieve a compensated adaptive equalizer control. In more specific embodiments, the adaptive control scheme is used with a ZF adaptation scheme in conjunction with a constant adaptation matrix. In other embodiments, the adaptive control scheme is used with a fast steepest-descent method using a variable adaptation matrix. In still other embodiments, the adaptive control scheme is used with a constant adaptation matrix, whereby a value of is statically calculated. If the adaptive control scheme is used with a decoupling matrix, a value of is statically calculated. An inverter is used between the data correlation matrix and the data-pattern compensation matrix.

The invention discloses a resolution controllable envelope generating operator-based multi-scale full-waveform inversion method. The method includes the following steps that: a resolution controllable resolution envelope generating operator is built, and the frequency band range of seismic data used for inversion is controlled by adjusting parameters; a full-waveform inversion method is built according to the resolution controllable resolution envelope generating operator, and the gradient of an objective function relative to model parameters and the calculation formula of iteration step size are inferred; and a steepest descent method is adopted to iteratively update the model parameters until algorithm conditions are satisfied. The method has better capacity of overcoming the problem of cycle skipping, is more suitable for the modeling of a lithologic oil and gas reservoir velocity model, and has more obvious advantages when seismic data are lack of low-frequency information or dominant frequencies in the seismic data are too high. With the resolution controllable envelope generating operator-based multi-scale full-waveform inversion method of the present invention adopted, the seismic data on which the resolution-controllable envelope generating operator has acted has enough low-frequency information regardless of whether the seismic data have sufficient low-frequency information or not; and the parameters can be adjusted, and multi-scale inversion can be realized.

The invention belongs to the technical field of modern traffic safety, and discloses a multi-index comprehensive evaluation method for the reliability of an urban rail train security detection sensor network. A series of indexes influencing the reliability of the urban rail train security detection sensor network is summarized, and the four indexes of time delay, the occupation ratio, the packet loss probability and the error rate are selected as evaluation indexes. The measured values of the evaluation indexes are normalized through a fuzzy mathematic theory, subordinative function values are obtained as samples, a BP neural network is built, the samples are trained, a weight value and a threshold value are modified through a steepest descent method, the samples are trained repeatedly in the same way until the error of an actual output value and the error of a calculated output value are within an acceptable range, the training is finished, and the obtained BP neural network with expert knowledge can comprehensively evaluate the reliability of the urban rail train security detection sensor network. The brand-new method is provided for evaluating the performance of the urban rail train security detection sensor network, and theoretical and practical supports are provided for optimizing the performance of the urban rail train security detection sensor network.

The invention relates to a single-end ranging method for an arc light high-resistance earth fault of a power transmission line. Sampling is carried out continuously on a fault phase current I phi, a zero-sequence current i0 and a fault phase voltage u phi(n) of a power transmission line, steady-state fault data sequences i phi (n), i0 (n), and u phi (n) after the fault are obtained, and a1(n) is calculated; a group of initial parameter solutions (1, UT, IS, and RF)<0> are given and a convergence error r(n) is calculated based on a formula; for a kth iterative solution (1, UT, IS, and RF) <k>, a steepest descent method is used and the solution to a fitting error minimal value is determined in a negative gradient direction as a searching direction; repeated iterative calculation is carried out continuously until an iterative solution (1, UT, IS, and RF)<k+1> of the (k+1)th times meets a given iterative termination principle:|(1, UT, IS, and RF)<k+1>-(1, UT, IS, and RF)<k>| < e, and an optimal solution (1, UT, IS, and RF)<*> is obtained, wherein the corresponding 1 expresses a fault distance. According to the method, on the basis of the characteristics of the fault arc dynamic nonlinearity and high resistance, a logarithm expression of an arc is provided according to the fault arc air discharge essence described by the Thompson principle. A defect that traditional arc differential equation calculation can not be used by the fault ranging algorithm directly can be overcome; and the practical flexibility is high.

In accordance with a particular embodiment of the present invention, a method is offered that includes providing a low-pass filter in an adaptive filter architecture employing a fast steepest descent method. The method further includes decomposing an error signal, injecting a small change to one or more weight parameters for a linear combiner, measuring changes in an error correlation vector, and calculating one or more gradients, wherein the gradients are processed by the low-pass filter. In more particular embodiments, one or more of the gradients are processed by the low-pass-filter in order to remove measurement noise and improve accuracy. In addition, a real gradient of the error correlation vector is monitored such that adaptations can be made due to non-linearity and non-constant characteristics of a channel. The low-pass filter may be replaced with a Kalman Filter for faster convergence.

The invention discloses a polypropylene melt index hybrid modeling method based on an integrated neural network. The method comprises steps: firstly, through acquiring key process variable data and melt index offline analysis data, preprocessing such as outlier detection and standardization is carried out on original data, and a training sample data set is built; then, a mechanism analysis method and a steepest descent method are adopted to build a polypropylene melt index simplified mechanism model, and a Bagging ensemble learning algorithm and an information entropy method are adopted to build a mechanism model prediction error compensation model based on an integrated BP neural network; and finally, the simplified mechanism model and the error compensation model are combined, the error compensation model is built for realizing online melt index estimation. Compared with current other modeling methods, the method of the invention has the advantages that the model generalization performance is improved; the polypropylene production process can be guided; and the polypropylene quality control can be effectively realized.

The invention provides a hybrid variation bioelectrical impedance imaging method, relating to the measurement on the conductivity of the pleural cavity of the body. According to the method, the potential spatial distribution condition is simulated by adopting a finite element method, a numerical model of the bioelectrical impedance imaging technology is established, a numerical model with large size is dispersed into small units, certain boundary conditions are applied and boundary voltage is calculated, a two-dimensional thoracic cavity numerical model and a three-dimensional thoracic cavity numerical model are respectively established, after the objective function of the hybrid variation algorithm is determined, the inverse problem solving is carried out by adopting the steepest descent method, the electrical impedance image reconstruction of the thoracic cavity is carried out, so that the resolution ratio of the reconstructed image is improved, therefore, the problems in the existing bioelectrical impedance imaging technology, the influences of noises and errors exist during the process of reestablishing the image with high resolution ratio, consequently, the excessive smooth and staircase effect are caused, and the accurate electrical impedance reconstructed image of the lung can not be obtained are solved.

The invention discloses an indoor positioning method based on a steepest descent method. The indoor positioning method is applied to positioning of indoor unknown nodes, position evaluation of the unknown nodes is carried out based on a signal intensity (RSSI) distance measurement method, optimization evaluation is carried out for the result through employing an optimization method to acquire the smallest-error position result. According to the method, under the condition of a lognormal shadow model, a reception signal intensity value is detected, influence of multiple factors onto an RSSI value is considered, a five-point quadratic curve fitting average method is employed, weight is carried out to correct the RSSI value, a least square method is utilized to correct a distance, the blind node position can be solved through a three-edge positioning algorithm, and the steepest descent method of the optimization method is employed to solve the accurate position of blind nodes. The method is advantaged in that indoor positioning precision is improved, and a positioning error is reduced.

The invention relates to a resistor-capacitor decoupling soft sensing method based on magnitude-phase characteristics, belonging to the technical field of the soft sensing technology of solution conductivity. The method is characterized in that the conductivity measurement in a conductivity cell is changed to the measurement of electrical property equivalent resistance-capacitance network of the conductivity cell, and the magnitude-phase characteristics of the resistance-capacitance network is estimated to obtain the resistance-capacitance parameter, wherein the resistance parameter is the resistance of the solution to be measured. Sine wave is used to excite the resistance-capacitance network; the steepest descent method is adopted for optimization and calculation based on the non-linear principle of least square; and multi-sampling of response signals is utilized to fit system function form and further obtain magnitude-phase characteristic parameter of the system. Then the magnitude-phase characteristics and the function relationship of the resistance-capacitance network are used to obtain the resistance-capacitance parameter. The invention has the effects and benefits of higher measurement accuracy and less iterations. In addition, the invention also provides a method for accurately obtaining the magnitude-phase characteristics of the linear system.

The invention discloses a fluorescence molecular tomography reconstruction algorithm based on an alternative iterative operation, which is characterized in that a weighted algebraic reconstruction technique and a steepest descent method are used alternately for solving. The fluorescence molecular tomography reconstruction algorithm comprises the following steps that (1), measurement data is acquired; (2), a linear relationship between the measurement data and target distribution is established; (3), a 2 norm minimization problem with a constraint condition is constructed; and (4), the weighted algebraic reconstruction technique and the steepest descent method are used alternately for solving the minimization problem, and a target distribution diagram is obtained. According to the fluorescence molecular tomography reconstruction algorithm, based on a light transmission theory and a finite element method, prior information such as an optical characteristic parameter and an anatomical structure is used, multipoint excitation and multipoint measurement are adopted, and the measurement data is obtained as far as possible, so that the pathosis of the problem is reduced; the weighted algebraic reconstruction technique and the steepest descent method are used alternately for solving the problem, so that a reconstruction result of fluorescence molecular tomography is improved effectively; and the fluorescence molecular tomography reconstruction algorithm has an important application value in the fields of molecular imaging, reconstruction algorithms and the like.

The invention provides a method for optimizing an attenuated phase-shifting mask (ATTPSM) of a nonideal photoetching system based on an Abbe vector imaging model. The method comprises the following steps of: setting transmittivity of different phase opening parts in the phase-shifting mask; setting a variable matrix Omega; setting a target function D as linear combination of an imaging evaluationfunction of an ideal image surface and an imaging evaluation function of an image surface of which the defocusing quantity is fnm; and guiding optimization on the pattern and the phase of the phasing-shifting mask by using the variable matrix Omega and the target function D. By using the vector imaging model and taking vector characteristic of an electromagnetic field into consideration during acquisition of a space image, the optimized mask is suitable for the photoetching system with small numerical aperture (NA) and also suitable for the photoetching system of which the NA is more than 0.6. By the method, the gradient information of optimizing the target function is utilized and a steepest descent method is combined to optimize the pattern of the phase-shifting mask, so the optimization efficiency is high.

The invention comprises a satellite-borne laser height measuring instrument footprint positioning method based on a passive target, solving the problem that echo signals of a flat terrain area are high in similarity by arranging an angle reflector array in the flat terrain area to mark the target echo to be measured in the flat terrain area, so that the high-precision positioning of the height measurement instrument footprint can be realized in a manner of waveform analysis in the flat terrain area, and the authenticity verification of the footprint position which is resolved by the height measurement instrument is finally realized. The method comprises angle reflector caliber design, angle reflector layout scheme, CCR energy contour line circle extraction, and footprint center extractionbased on a steepest descent method. The system includes a CCR caliber calculation module, a CCR layout scheme module and a CCR-based footprint positioning module. According to the technical scheme, the positioning of the height measurement instrument footprint can be achieved through the waveform analysis mode in the flat terrain area.

The invention provides an optimization method of optical proximity effect correction based on an Abbe vector imaging model; according to the method, the transmittance of an opening part and a light-masking part of a mask is set; a variable matrix omega is set; an object function D is constructed to be the square of an euler distance between a target pattern and an image in a photoresist corresponding to the current mask; the optimization of the mask pattern is guided by the variable matrix omega and the object function D. The mask optimized by the invention is not only applicable to cases with a less NA, but also applicable to cases with a NA higher than 0.6. Meanwhile, by using gradient information of the optimized object function, and combining a steepest descent method, the invention optimizes the mask pattern with high optimization efficiency.

In a method for optimizing a codebook including a quantized channel between a base station transmitter and a portable terminal: a) select the codebook to be optimized; b) a steepest descent algorithm is applied to respective codewords of the codebook selected in a), wherein the codewords are transformed so that a correlation between the codewords is minimized, and a function indicating the correlation between the codewords is differentiated and is multiplied by a predetermined step size and is used for a parameter for generating a new codeword in the steepest descent algorithm; c) it is determined whether the codebook including the codewords transformed in b) is optimized; and d) the codebook is stored when it is determined in c) that the codebook is optimized.