49 results about "Minimum norm" patented technology

Disclosed is a shaft sleeve part surface defect on-line detection method based on compressed sensing. Compressed sensing description of a part surface defect image is built through a machine vision and compressed sensing method, and an optical imaging and defect detection model highlighting surface defects is built; a part sample image of typical defects is collected, after denoising and necessary image preprocessing are carried out, sampling frequency adjustment and size normalization are carried out, a sample is trained and a redundant dictionary is built; a proper orthogonal basis decomposition matrix and a random observation matrix are designed, a combined orthogonal matching pursuit algorithm is selected, solution of the minimum norm l0 is converted into the problem of solving the optimal solution to reconstruct a defect image, spare representation of the image to be detected is calculated, and defect recognition is carried out on a part to be detected according to built judgment and recognition standards. An on-line detection system with the functions of feeding, positioning and adjustment, image collection, image processing, defect detection and recognition, part separation and the like is built, and rapid detection on the surface defects of the shaft sleeve part is achieved.

The invention discloses a method for inversing a wavenumber domain of underground three-dimensional magnetic distribution by magnetic total tensor gradient and an application thereof. The method comprises the steps: known prior information of a study area and observation magnetic total tensor gradient anomality are read; inversion parameters, an inversion depth range, step length, and the like areset; constrained inversion is performed on the observation magnetic total tensor gradient anomality to obtain a magnetic three-dimensional mesh model; forward is performed on the observation magnetictotal tensor gradient anomality to obtain theoretical magnetic total tensor gradient anomality; subtraction is performed on the theoretical anomaly and the observation anomaly to obtain a theoreticalanomaly residual error; and if a minimum norm solution of the anomaly residual error does not satisfy limits of error, the inversion is performed on the anomaly residual error to obtain an anomaly residual error model, the magnetic three-dimensional mesh model is updated, the steps are repeated until the anomaly residual error satisfies requirements, and a final magnetic three-dimensional mesh model is an inversion solution. The method can reflect three-dimensional distribution of magnetic properties (magnetic susceptibility or magnetization intensity) of underground space geologic bodies, and has the application value for improving the three-dimensional inversion efficiency and fine geological simulation of large data bodies.

The invention relates to a PSO (Particle Swarm Optimization) extremity learning machine based strip steel exit thickness predicting method, which basically comprises the steps below: 1) analyzing a strip steel data signal by utilizing a data processing software, selecting four parameters which greatly influence the thickness of the strip steel exit and includes a roll force, a roll gap, a roll speed and a motor current, and inputting the four parameters as input variables into an extremity learning machine in the prediction of the thickness of the strip steel exit; 2) performing selective optimization on parameter input weights and a hidden layer offset value in the extremity learning machine by using the PSO, analyzing and determining output weights by applying a generalized inverse way to obtain an output weight matrix with a minimum norm value in the extremity learning machine, and accordingly obtain optimal parameters of the extremity learning machine; 3) modeling the obtained optimal extremity learning machine; 4) predicting the thickness of the strip steel exit by inputting the four parameters in the step 1) into the optimized extremity learning machine. By applying the PSO extremity learning machine based strip steel exit thickness predicting method, analysis aiming at the rolling production process is carried out, the prediction for the thickness of a rolled piece exit is performed, relevant technical parameters affecting the quality of the strip steel are further analyzed, and real-time control and adjustment for the rolling production process are further carried out.

A noise suppression apparatus selectively uses an adaptive beamformer and fixed beamformer for each frequency. A direction of a null of the fixed beamformer is determined from a direction of a null automatically formed by the adaptive beamformer. Filter coefficients of the adaptive beamformer based on an output power minimization rule are calculated by a minimum norm method using a norm of the filter coefficients as a constraint. The above selection is made based on, for example, a depth of a null automatically formed by the adaptive beamformer in the selection.

Systems and methods are described using a Hermetic Transform and inverse Hermetic Transform for transmitting and receiving a signal with multiple subcarriers, oversampling the received signal, and providing the resulting signals to a Hermetic Transform to provide a higher resolution output signal. An inverse Hermetic Transform is used to process data for transmission. The result is comparable to a form of OFDM that has orthogonality in a minimum norm / least-squares sense.

The invention discloses a method for measuring theoretical life of a mechanical product. The method includes evaluating an optimum confidence interval of reliability in Weibull distribution of three parameters on the basis of life test data of samples by means of self-service norm weighting method, establishing the optimum parameter information vector of the Weibull distribution by six minimum norm weighting standards on the basis of empirical value and theoretical value of reliability, subjecting the optimum parameter information vector to self service and sampling, generating parameter information vector, solving estimated true values of parameters and confidence intervals of the estimated true values under condition of giving the confidence level, setting up an estimated true value function and an optimum confidence interval function of the reliability according to the estimated true values and the confidence intervals, and solving the theoretical life of the mechanical product. A plurality of different norms are optimized and select to process original life data, and the optimum information vector of the Weibull distribution of a plurality of side characteristics can be displayed integrally, and measurement of the theoretical life of the product is optimized after twice optimization.

The invention discloses a cancer gene expression profile data identification method based on integration of extreme learning machines. The method includes the steps of selection and integration of member extreme learning machines (ELMs). The method concretely includes the steps that preprocessing is carried out on a cancer gene expression profile data set, wherein the preprocessing includes gene selection and normalization of expression profile data; N sample sets are generated through a Bagging method, and each sample set is divided into a training set and a verification set according to a certain proportion; N ELMs are generated on the N training sets in a learning mode, and L ELMs (L&1t; N) with the highest recognition rate on the corresponding verification sets are selected to form an alternative member ELM base; K member ELMs (K&1t; L) forming an integrated system are selected from L ELMs based on the particle swarm optimization algorithm; the integrated vote weight of K member ELMs is worked out by utilizing the minimum-norm least square method; an integrated ELM system is obtained, and the integrated ELM system is used for performing tumor recognition on a newly increased cancer gene expression profile sample. Through the method, the cancer gene expression profile data can be quickly and accurately recognized.

The invention discloses a method and system for positioning short-circuit faults of an intelligent distribution network considering load influences. The method comprises the following steps of: obtaining a current phasor at each power source in the distribution network before and after the fault, and calculating a short circuit current provided by multiple power sources in the distribution networkwhen the fault occurs; obtaining synchronous current data and voltage data of each end point of each section before and after the fault, and calculating a current level of the internal load of each section; calculating a current phasor characteristic value of each section when the fault occurs according to the current level of the internal load of each section and the short circuit current when the fault occurs; collecting the current phasor characteristic values of each section under multiple time scales, and constructing a current space-time characteristic matrix of each section; calculating a norm value of the current space-time characteristic matrix of each section, and storing to a characteristic value unit; and traversing the characteristic value unit to find a minimum norm value, and the section corresponding to the minimum norm value is a fault section.

The invention provides a super-resolution imaging method and device based on sparse matrix reconstruction and electronic equipment. The super-resolution imaging method comprises the steps of obtainingan echo signal of a target object, obtaining a first expression according to the echo signal, transforming the first expression, and obtaining a two-dimensional sparse matrix of the echo signal; by obtaining the two-dimensional sparse matrix of a SOONE function, transforming the two-dimensional sparse matrix, obtaining a minimum norm solution, conducting iterative optimization on the minimum normsolution, and obtaining the minimum value of the two-dimensional expression of the SOONE function on solution space; projecting the minimum value into the solution space, and obtaining a reconstruction signal; processing the reconstruction signal through a compressed sensing short aperture imaging algorithm, and obtaining a target imaging result. By means of the method, the problem is solved thatwhen the target movement is complicated, a high-resolution ISAR image is difficultly obtained by using a RD algorithm.

The invention relates to a plant classification method based on sparse expression dictionary learning. The method is technically characterized by comprising the following steps: performing parameter initialization, wherein the parameter initialization comprises arranging the size, a sparse limitation factor and an error tolerance parameter of each plant type dictionary; for a training sample of each type of plant blade images, obtaining an over-complete dictionary of each type of blade images by use of a K-SVD algorithm; splicing the over-complete dictionary of each type of blade images after training to form a redundancy dictionary, and performing normalization processing on each column of the redundancy dictionary; obtaining a sparse coefficient through solving a minimum norm; and calculating residual errors and selecting a corresponding sample type with a minimum difference as a final identification result of a sample to be identified. According to the invention, the over-complete dictionaries are solved by use of type-based dictionary learning and sparse representation of an image to be identified is calculated, such that the calculation time of an algorithm is reduced, the requirement for real-time performance is met, the obtained identification rate is quite high, and the average identification rate is as high as more than 95%.

The invention discloses a rail train line-scan digital camera image distortion correction method, which comprises the following steps of: taking a distortionless train line-scan digital camera image of the same model as a standard template, and respectively performing image correction on each single carriage or one section of the single carriage of a train; subjecting the image to oversampling processing before image distortion correction is carried out, so that image information loss possibly caused when the image contour is extracted in the next step can be avoided; extracting the image contour before the minimum norm optimization problem is established so that the possible influence of illumination difference, dust, water stain, stain and random noise interference between the standard template image and the distorted image can be effectively solved, correction is enabled to be more accurate and the reliability is enabled to be higher. According to the method, a linear equation problem is converted into a constraint problem of a minimum norm, and solving is carried out through an optimization method, so that a more stable and reliable reduction matrix can be obtained, and a corrected result is more stable and reliable.

The invention discloses a neural-network-based rapid compensation method for photoelectric encoder. The method can increase compensation precision and compensation speed, and can simplify compensation procedures and reduce requirements to a needed instrument. The method includes: first, randomly rotating the photoelectric encoder in a circle, and acquiring a training sample; then adopting a single input and single output three-layer feed-forward Fourier neural network to establish a photoelectric encoder error compensation model, and estimating the number of nodes in hidden layers based on the characteristics of an object function; at last based on the actual result, correcting the number of the nodes so as to prevent the inefficiency due to repeated and varied attempts in a trial-and-error method. According to the invention, directed to the problem of slow convergence of the iteration training method, the method, through mapping the errors to nodes in the hidden layers of orthogonal triangle function basis, rapidly solves a minimum norm solution from the hidden layers to a weight of an output layer, which substantially reduces the time required by solving the weight of the neural network and greatly reduces time complexity and space complexity in calculation.

The invention relates to a modulation transfer function (MTF) in-orbit measuring method for a gazing satellite area array CCD camera. The method is characterized in that an end-to-end point spread function model is established between point-source distribution and image distribution, point source is designed and laid; point spread function value of each image element in a point source zone is calculated, and then the point spread function parameter of the area array CCD camera can be obtained by using minimum norm solution, furthermore an optical transfer function can be obtained; finally, the optical transfer function is delivered and is the modulation transfer function, a first modulation transfer function value is taken as a reference for normalization processing, the modulation transfer function on any space direction theta can be obtained through transformation of coordinates; and the modulation transfer function at Nyquist frequency on any space direction theta is taken as a MTF index evaluate value of the gazing satellite area array CCD camera. The in-orbit measuring method solves the technical problem of MTF no in-orbit measuring of the gazing satellite area array CCD camera in background technology.

A novel magnetic resonance imaging method is described, wherein undersampled magnetic resonance signals are acquired by a receiver antenna system having spatial sensitivity profiles and the image being reconstructed from the undersampled magnetic resonance signals and the spatial sensitivity profiles. The reconstruction of the image is provided by an optimization of a cost function which accounts for any of noise statistics, signal statistics, and the spatial response function, the latter of which is defined by the spatial signal response from the object to be imaged, separately for each individual pixel.

The invention discloses a two-dimension phase fast unwrapping method using an L<0> norm cost function; the method comprises the following steps: numerical realization of solving a differential equation algorithm, firstly converting a differential relation into a difference relation corresponding to discrete data, then carrying out difference equation iteration solving, wherein each step treatmentis carried out on the basis of the previous step treatment result in the iteration process, and the difference [delta] [phi] between the two processing steps becomes the solving important content. Thetwo-dimension phase fast unwrapping method using the L<0> norm cost function can solve the problems that an existing phase unwrapping method is low in efficiency when processing big data blocks. Themethod aims at the conventional minimum norm unwrapping method, analyzes the cost function characteristics, and provides the two-dimension phase fast unwrapping method complying with the L<0> norm cost function; combined with a data partitioning scheme, the method can solve the full low frequency aberration convergence problems in the linear differential equation, and improving the unwrapping efficiency.

A compound-weight time-series data calculator creates a compound spectrum matrix based on standard mass spectra of various compounds stored in a standard mass spectrum library. Then, the calculator determines, in the form of a linear regression model, the relationship of the matrix, a measured vector based on the data acquired at one measurement time point, and a compound weight vector at the measurement time point, and estimates the unknown weight vector by a minimum norm estimation in which a regularizer is introduced. A compound-weight time-series graph creator creates a compound-weight time-series graph showing a temporal change in weight for each target compound, based on the compound weight vector obtained at the measurement time points within a specified time range. A peak which appears on this graph is used for the quantitative determination of the compound or determination on the presence or absence of the compound.

The invention relates to a method based on radial basis network algorithm for acquiring yield of integrated circuits, belonging to the technical field of integrated circuits. The method includes the steps of: according to the technological parameters provided by an integrated circuit manufacturer and by adopting radial basis network algorithm, establishing a substitute model for replacing the circuit emulations by taking the technological parameters as the independent variables of the substitute model and the performance indexes of the circuits as the function values of the substitute model; acquiring a technological float value which is most easy for disabling the integrated circuits according to a minimum norm method; conducting systematic sampling on the acquired substitute model and the technological float value most easy for disabling the integrated circuits to acquire the sampling points and the performance indexes of the circuits; and acquiring the yield of the integrated circuits through a statistical approach according to the sampling points and the performance indexes of the circuits. The method has the advantages that the circuit emulation times in the process of acquiring the yield of the integrated circuits can be reduced, the time spent on analyzing the yield of the integrated circuits is reduced, the design cycle of the integrated circuits is shortened, the production of the integrated circuits is sped up, the cost of the integrated circuits is reduced and the economic value is increased.

Known chip equalizers for Wideband-Code Division Multiple Access (W-CDMA) employ a co-efficient calculator that implements a Minimum Mean Square Error (MMSE) solution to a least squares technique for obtaining equalizer coefficients in response to receipt of a pilot sequence. However, this solution results in an undesirably high processing overhead in downlink receivers operating in a W-CDMA communications system. Consequently, the present invention provides a computationally simpler technique to calculate equalizer coefficients by implementing a minimum-norm solution to the least squares problem.

A processing method of periodic errors in an inertia device is provided in the present invention. The parameters of frequency, initial phase, etc. of remarkable periodic errors are demarcated by the initial output data sampling of the device, and the online compensation in real time is performed in the subsequent time to improve the precision of the device. Wherein, the frequency demarcating process includes an initial demarcating process and a precise demarcating process. The initial demarcating determines the frequency of the periodic error items by a PSD analyzing method. The precise demarcating determines the frequency by transforming the questions into the minimum norm local optimization problems which have nonlinear multi-restriction in the optimization theory for modeling and the design of the arithmetic. The performing of the method is quick and exact and has a high precision without influencing the bandwidth and useful signals. The method is especially adequate for the periodic error processing of the high-frequency and over-sampling data of the inertia device and is also adequate for the processing of data including periodic errors in other fields.

The invention discloses a power grid fault diagnosis method based on a multi-variate power grid database. The power grid fault diagnosis method based on the multi-variate power grid database comprisesthe following steps that fault data of a multi-element power grid are acquired, and a fault sample training characteristic matrix is established according to the fault data; a sparse coefficient x isobtained through test data Y to be diagnosed obtained by the minimum norm constraint; and according to the sparse coefficient x, test sample data are obtained, the minimum residual between the obtained test sample data and each fault sample characteristic data in the fault sample training characteristic matrix is obtained by calculation, comparing is carried out to obtain a minimum value of the minimum residual, a fault category is determined, and the power grid fault diagnosis is completed. According to the power grid fault diagnosis method, the accuracy of fault diagnosis can be improved, and the fault false alarm rate is reduced.

The present invention provides a cardiac outline reconstruction method and a magneto cardiogram measuring system. The cardiac outline reconstruction method comprises the steps of: obtaining a first cardiac outline from power of source current vector (source current power) obtained using an array-gain constraint minimum-norm with recursively updated gram matrix; AGMN-RUG) spatial filter; obtaining a second cardiac outline by applying a coherence mapping method; and configuring a third cardiac outline by coupling the first cardiac outline and the second cardiac outline with each other.

The invention discloses a compression method serving industrial interconnection data acquisition and storage. The method belongs to the field of data compression, and comprises the following steps: S1, defining a range value, a range percentage, a longest interval and a shortest interval on a platform, S2, carrying out analysis processing according to a process of judging the longest interval, carrying out range calculation and judging the shortest interval, and S3, storing data subjected to the processing process. According to the method, dynamic configuration is achieved technically, the compression hit rate is increased reasonably, screening of feature data is refined, the operand is reduced to the minimum standard, and it is guaranteed that the operation efficiency is the highest; resource controllability is economically achieved, free configuration is achieved according to requirements, and rationality and integrity of data are considered while resource use is reduced as much as possible; dynamic adaptation is achieved in performance, horizontal expansion is achieved through a distributed architecture, dynamic adaptation is conducted according to different data scales, and thecollection and storage performance is greatly improved.

The invention provides a partial reconfiguration based broadband spectrum perception method, aiming to apply the idea of partial reconfiguration to a minimum norm solution based interior point method. Different from a traditional reconfiguration method without adjustment, the broadband spectrum perception method is characterized in that gradual adjustment is performed according to sampling values and through multiple iterative calculations. The number of iterations is regular but changes with the energy of reconfigured sub-channels. If the energy is large, the number of the iterations is reduced, and if the energy is small, the number of the iterations is increased. The partial reconfiguration based broadband spectrum perception method can be used in the situation that the sparseness is unknown or even in the broadband spectrum scenario with the changed sparseness, calculation time is reduced under the condition that detection precision is guaranteed, and timeliness of detection is improved.

The invention discloses a parallel compression sensing method for multi-channel synchronous acquisition of heart sound signals. The method firstly includes , according to the time-frequency domain properties of multi-channel heart sound signals collected synchronously, selecting one main signal and analyzing the correlation between other signals and the main signal; then compressing the main signal by using the test matrix, and converting the problem of recovering the original signal from the observation matrix into the problem of solving the minimum norm to realize the reconstruction of the main signal; and reconstructing the main signals and finally reconstructing other signals according to the correlation between the main signal and other signals, so as to realize the parallel compressed sensing of multi-channel heart sound signals. By processing the main signal, the method realizes the parallel compression sensing of the multi-channel heart sound signals, greatly improves the reconstruction and feature extraction speed and accuracy of the multi-channel heart sound signals, realizes the long-term monitoring and storage of the heart sound signals, and improves the reliability ofvarious predictions based on the heart sound signals. The invention can guarantee the wearable equipment to run for an ultra long time under the energy-saving state, and further can promote the application of the parallel compression sensing processing technology to other biological signals collected synchronously in multiple channels.

The invention relates to the technical field of laser and temperature field measurement, and puts forward a temperature field reconstruction method which is easy in implementation. The reconstructionstep is simple, operation efficiency is higher, and temperature anomaly can be detected in real time. In addition, the invention adopts the technical scheme that the optimization and reconstruction device and method for a temperature field on the basis of a minimum norm function utilizes a computer to control a laser to alternately generate two frequency sweep laser pulse signals of different frequency bands, after the laser signal passes through a heat source to be detected, a sensor receives the signal and deduces laser absorbance, and then, temperature field reconstruction is carried out bythe obtained laser absorbance. The method is mainly applied to a temperature field measurement occasion.

The invention discloses a target point allocating method for robot team formation. The method comprises the steps of 1) calculating a path length set of each robot to all target points according to position coordinates of each robot and the target points; 2) selecting elements from the path length sets of the robots according to a permutation and combination random drawing method to form a plurality of formation matrixes; 3) calculating the norm of each formation matrix and sorting, wherein the formation matrix corresponding to the minimum norm serves as the matrix with the shortest path of the team formation; and 4) judging whether the matrix with the shortest path of the team formation is the final target point allocating matrix or not and selecting the final target point allocating matrix to complete the target point allocating of the robot team formation. The method disclosed by the invention has the beneficial effects that while the obstacle avoidance can be realized, the high-efficiency team formation with the least time and path consumption, i.e., the team formation initialization, can be realized at the same time in an environment with obstacles, thereby achieving the optimal path planning.

A method of identifying an epileptogenic zone of a subjects brain includes: receiving a plurality N of physiological brain signals that extend over a duration, each of the plurality N of physiological brain signals acquired from the subjects brain; calculating within a time window a state transition matrix based on at least a portion of each of the plurality N of physiological brain signals, wherein the state transition matrix is a linear time invariant model of a network of N nodes corresponding to the plurality N of physiological brain signals; calculating a minimum norm of a perturbation on the state transition matrix that causes the network to transition from a stable state to an unstable state; and assigning a fragility metric to each of the plurality N of physiological brain signals based on the minimum norm of the perturbation for that physiological brain signal.

The invention relates to the technical field of electric power, in particular to a power signal disturbance analysis method based on TLS-ESPRIT. The method includes: inter-harmonic detection, Simulation experiment, an actual data analysis and conclusion step. Harmonic inter-harmonic detection comprises construction of a signal mathematical model and subspace extraction, the subspace extraction interferes with a signal subspace through disturbance signals with the minimum norm, the influence of noise in the signal subspace is reduced, the subspace extraction can correct noise existing in the two subspaces at the same time, and the calculation accuracy is further improved; The simulation experiment comprises simulation signal calculation and matlab simulation environment simulation. The method combines the total least square method and the subspace rotation invariance to realize the detection of the power signal disturbance, and the simulation analysis proves that the algorithm is accurate and effective, the realization process is simple, and the calculation amount is small. The feasibility of the algorithm is proved through analysis of actual data, and the method has an applicationprospect in actual power signal interference processing.