41 results about "Gaussian radial basis function" patented technology

The invention belongs to the field of mining safety and relates to a rock-burst acoustic emission predicting method based on a support vector machine (SVM). The rock-burst acoustic emission predicting method includes using SVM regression prediction as a means, selecting energy release value as judgment indicator for predicting rock burst on the basis of test research analysis, using burst tendency physical and mechanic parameters and acoustic emission signal feature parameters as influence factor of rock burst, and generating SVM input vector by extracting sample data according to these parameters; selecting Gauss radial basis function (RBF) as kernel function of the SVM in setup of a SVM prediction model, and using the K-CV cross validation algorithm to optimize and select the optimal penalty parameter C and kernel function parameter g; applying a fuzzy information granulating data processing method into the regression prediction of the SVM so as to predict change trends and change space of the predicted values of the rock burst.

The invention relates to a method for predicting performance indexes in a polyester fiber spinning process, in particular to a method for predicting performance indexes in a polyester fiber spinning process based on a least square support vector machine (LS-SVM) optimized by a particle swarm optimization (PSO) algorithm improved by a power law. The method comprises the following steps of selecting production parameters, including a spinning speed, a spinning temperature, an air blowing temperature and an air blowing speed, in the polyester fiber spinning process as feature information, performing linear function normalization, and establishing an input sample dataset; determining main performance indexes, including half-fold elongation rate, irregularity of the half-fold elongation rate, breaking strength and elongation capability, influencing polyester fiber quality, performing logarithmic function normalization, and establishing an output sample dataset; and building an LS-SVM model according to the input and output sample datasets, adopting a Gauss radial basis function (RBF) as a kernel function of the LS-SVM, and selecting an optimal penalty factor C and a kernel function parameter sigma by using PSO. The PSO process is improved according to the power law, so that the optimization speed can be greatly increased and accurate prediction is realized.

The invention discloses a prediction control method and system for a magnetic levitation system. By historical data of the magnetic levitation ball system and through adopting a system identificationmethod, an autoregressive model which takes a Gaussian radial basis function network as a coefficient and has an exogenous variable is established to describe nonlinear dynamic characteristics betweenan input voltage of an electromagnetic winding and a position of a steel ball. An ARX model coefficient is fit by an RBF neural network, so that an RBF-ARX model can well depict the nonlinear dynamiccharacteristics of the magnetic levitation ball system. Then, on the basis of a principle of a predictive functional control algorithm, a structure of a control input is determined so as to completerolling optimization and error correction. The predictive functional control algorithm has higher tracking capacity and higher robustness than PID control, and has a smaller online calculation amountand a higher control speed than conventional model prediction control; and when a system set value is a step signal, the control method can implement unbiased tracking.

The invention relates to a power load prediction correction method, especially to an improved-support-vector-machine-based power load prediction correction method. With the method, problems that the current and future power demand changes and the power demand development trend can not be reflected accurately and the power load accuracy prediction is low can be solved. Therefore, the invention puts forward an improved-support-vector-machine-based power load prediction correction method. The method comprises: step one, sample data are selected from a database or a field; step two, the sample data are coded; step three, pretreatment is carried out on the coded sample data; step four, a kernel function is initialized; and the sample data after pretreatment are processed by selecting a Gaussian radial basis function; and step five, power load prediction is carried out by using an improved support vector machine. The method is applied to the power load prediction correction field.

The invention provides methods for constructing and predicting the leaf trait of a woody plant and a photosynthetic characteristic model based on a DNA methylation level, and belongs to the technicalfield of biological analysis. The predicting method comprises selecting important characteristic variable embodying a geographic position difference based on a random forest, screening out 7 leaf characteristic variables, determining an optimal cluster number, and obtaining each group of cluster leaf samples by using an improved FCM clustering algorithm; according to the correlation between variables and the importance of Enzyme digestion combination obtained by a gradient boosted tree, obtaining an important enzyme digestion combination in each group of cluster leaf samples; by using the DNAmethylation level of the enzyme digestion combination as a regression variable, constructing LS-SVM regression prediction model based on Gaussian radial basis function; inputting the DNA methylation level of important enzyme digestion combination to accurately predict a leaf shape factor, leaf area and a net photosynthetic rate. The method is used for predicting the phenotypic characteristic and the photosynthetic characteristic of the woody plant, and screening individuals of woody plants with excellent traits.

The invention provides an improved-Fisher-based chemical process fault diagnosis method. The method comprises: step one, original chemical process fault diagnosis are collected and normalization processing is carried out on the data, wherein the data are classified into a training set and a testing set; step two, the training seat is inputted into an improved class-separation-distance kernel Fisher method, a threshold parameter is outputted, and a parameter of a Gaussian radial basis function is selected optimally by using a cross validation method; step three, the testing set outputted at the step one is inputted into the improved class-separation-distance kernel Fisher method to carry out projection; and step four, according to the threshold parameter outputted at the step two, whether the data are boundary points after projection at the step three is determined and a fault type is determined by combining an improved K-NN algorithm based on a mahalanobis distance. According to the method, distribution of sampling data in projection space is improved; and with introduction of a boundary threshold parameter and combination of the mahalanobis distance and the improved K-NN algorithm, the classification accuracy of the total samples is increased under the circumstance that the classification time is minimized.

The invention belongs to the crossed technical field of computer mode identification and nano material, and relates to a low-dimensional nano material identification method based on an SEM image. The method comprises the following steps of: (1) preprocessing a known nano material SEM image sample; (2) performing two-dimensional wavelet transformation on the preprocessed image to get sub-image matrixes on different frequency bands; (3) extracting characteristics of the sub-image matrixes on each frequency band, and taking a statistical value of each sub-image matrix as a characteristic value for representing surface texture of the nano material; (4) according to the characteristic value, taking a Gaussian radial basis function as a support vector machine kernel function to find an optimal hyperplane between any two classes, and creating a classification model for different classes of nano materials; (5) extracting a texture characteristic value of the known nano material SEM image sample, and identifying the unknown nano material by voting according to the classification model obtained in the step (4). The low-dimensional nano material identification method based on the SEM image represents and distinguishes different nano material structure types more accurately and effectively, and has the advantages of high accuracy, strong expansibility, high degree of automation and the like.

The invention provides a method and a device for measuring harmonic waves and inter-harmonic waves in a power system. The method comprises the following steps of: a Gaussian window is added for the time domain of the sampled electric signal in the power system and fast discrete Fourier transform is performed; a signal spectrum is generated, a plurality of Gaussian radial basis functions are constructed according to the number of wave crests in the frequency spectrum; the signal frequency spectrum is processed, iterative training is carried out on the to-be-trained target function, so that theto-be-trained target function approaches the processed signal frequency spectrum, and the harmonic signal and the inter-harmonic signal are respectively measured according to the parameters corresponding to the trained target function. Thus, according to the embodiment of the invention, a series of Gaussian radial basis functions are constructed according to the number of wave crests of the frequency spectrum after fast Fourier transform, the frequency spectrum generating the inter-spectral interference is approximated by utilizing linear superposition of the Gaussian radial basis functions, harmonic and inter-harmonic components generating the main lobe interference in the electric signal can be accurately distinguished, and the measurement precision is relatively high.

The invention relates to a method for optimizing a multi-kernel multi-feature fusion support vector machine and identifying a bearing fault. The method comprises a step of selecting bearing vibrationsignals collected under a single sensor, a step of decomposing bearing vibration signals at different rotational speeds by EMD to obtain IMF energy entropy and IMF permutation entropy, a step of extracting IMF energy entropy and IMF permutation entropy at different rotation speeds and fusing the IMF energy entropy and IMF permutation entropy to obtain fusion features including different rotationalspeed information for support vector machine training samples so as to obtain the multi-kernel multi-feature fusion support vector machine which is adapted to fault identification at different rotation speeds, a step of integrating Gaussian radial basis function kernel and polynomial function kernel performance, allowing the training samples to be in linear regression from a nonlinear function space to high-dimensional space mapping such that the training samples are classified according to different characteristics, forming a multi-kernel least square support vector machine, and enabling thesupport vector machine to identify a fault feature under a variable load, and a step of carrying out parameter optimization on the training samples with a self-adjusting particle swarm algorithm withstrong convergence, comparing the training samples and a test sample, and identifying the bearing fault.

The invention provides a Chinese text classification method. The method includes the following steps: (1) text preprocessing, (2) feature representation and feature extraction, (3) classifier design,and (4) performance indicators. According to the method, a new RBF neural-network algorithm is employed, K-means is used to derive a center point and width of hidden items through Gaussian radial basis function, output results obtained by a hidden layer are combined, thus a classification result is obtained, values of an accuracy rate, a recall rate and F measurement of the algorithm are very high, and a classification effect is good.

The invention relates to a method for predicting a coagulation bath link performance index in a carbon fiber precursor production process and especially relates to a method for predicting a coagulation bath link performance index in a carbon fiber precursor production process of a least square support vector machine (LS-SVM) optimized by particle swarm optimization (PSO) improved based on a power-law rule. The method comprises the following steps: a fiber initial radius, a coagulation bath effective length, an initial speed and a pulling speed of a coagulation bath link in the carbon fiber precursor production process are selected as feature information, and an input sample data set is established after linear function normalization is performed; major performance indexes influencing carbon fiber quality, i.e., a fiber final radius and a coagulation layer thickness are determined, and an output sample data set is established after logarithm function normalization is performed; and according to the input-output sample data sets, establishing an LS-SVM model, by taking a Gaussian radial basis function (RBF) as a nucleus function of the LS-SVM, an optimal punishment factor C and a nucleus function parameter sigma are selected by use of the PSO; and the searching optimization process of the PSO is improved according to the power-law rule such that the searching optimization speed can be greatly improved and accurate prediction is realized.

The invention relates to a method for predicting performance indexes of a drafting link in a carbon fiber precursor production process, in particular to a method for predicting performance indexes of a drafting link in a carbon fiber precursor production process based on a least square support vector machine (LS-SVM) optimized by a particle swarm optimization (PSO) algorithm improved by a power law. The method comprises the following steps of selecting a six-stage draft ratio of the drafting link in the carbon fiber precursor production process as feature information, performing linear function normalization, and establishing an input sample dataset; determining main performance indexes, including linear density, precursor strength and breaking elongation rate, influencing carbon fiber quality, performing logarithmic function normalization, and establishing an output sample dataset; and building an LS-SVM model according to the input and output sample datasets, adopting a Gauss radial basis function (RBF) as a kernel function of the LS-SVM, and selecting an optimal penalty factor C and a kernel function parameter sigma by using PSO. The PSO process is improved according to the power law, so that the optimization speed can be greatly increased and accurate prediction is realized.

The invention discloses a networked numerical control system man-in-the-middle attack detection method based on a Gaussian radial basis function classifier, and the method comprises the steps: firstly, analyzing the physical sensor information of a numerical control system from the perspective of a control level; then, starting from the perspective of data driving, considering the data of the industrial control system under the offline condition as normal working condition data, and further analyzing the working condition data under the online condition by taking the normal working condition data as a reference; taking the deviation degree between the working condition data and the historical working condition data under the online condition as sample data; then, starting from the perspective of statistics, taking the mean value and the variance as feature information of the sample data; and finally, starting from the perspective of a machine learning method, training a support vectormachine based on a Gaussian radial basis function to obtain a classifier, and achieving correct classification of attacks. The method can solve the problem that a traditional IT security detection scheme is difficult to detect professional attacks; and the theoretical framework is simple, easy to implement and good in generalization capability.

The invention discloses an imaging system design method based on a Gaussian radial basis function curved surface. An optical system with a field of view smaller than a required value serves as a design starting point, then the field of view is gradually enlarged, based on the thought of field of view expansion and Gaussian continuation, new Gaussian functions are sequentially added to the edge of a curved surface along with expansion of the field of view, the size of the curved surface is gradually expanded, and an imaging system finally meeting design indexes is obtained through progressive optimization. The method is simple, easy to implement and particularly suitable for design of a large-field-of-view imaging optical system. According to the method, convenience is provided for an actual imaging system to apply the Gaussian radial basis function curved surface, the method is convenient, easy and convenient to use and high in applicability, and for free-curved surface imaging systems of various applications and various system structures, the method can be used for gradually completing system design from a simple system as a starting point.

The invention discloses a support vector machine based automatic focusing method of ultrasonic phased arrays of ring welds. The method includes: establishing a rectangular coordinate system, taking a least squares support vector regression (LSSVR) machine as the regression model and Gaussian radial basis function as the kernel function, performing training with input and output which are subjected to standardized processing, and establishing a prediction model of LSSVR initial function; taking a generalization error calculated with a k-fold cross validation method as the target function, optimizing the LSSVR initial function by adopting an optimization method combining a coupling-simulated annealing algorithm with a grid searching method to acquire an optimal hype-parameter and an LSSVR optimization model, predicting tested data to acquire the output, performing inverse normalization processing to obtain the optimal acoustic beam path plan of a corresponding partition, performing reverse solution according to the Fermat theorem to obtain time delay, and applying ultrasonic signals to perform focusing detection on ring weld defects. By the arrangement, the method has the advantages that intelligent ring weld defect detection is realized, and detection precision and efficiency are effectively improved.

The invention belongs to the field of information processing, and particularly relates to a parameter determining method of a Gaussian radial basis function agent model. According to the technical scheme, the parameter determining method includes the steps that sample space is linearly mapped into a unit cube; the local densities of sample points are calculated according to sample distribution conditions; the minimum distance between the sample point with the minimum local density and other sample points is calculated; the kernel widths of the sample points are determined; the weight coefficients of basis functions corresponding to the sample points are determined. The parameter determining method has the advantages that the method is clear in logic, easy and convenient to operate and easy to execute; the calculated amount is hardly increased with the basis-function kernel width determining method; the Gaussian radial basis function agent model obtained with the method can be commonly used for even/uneven samples, and the parameter determining method is reliable, efficient and high in accuracy.

The invention discloses an IFOA-SVM-based high-voltage switch cabinet state evaluation method. The method comprises the steps: firstly, collecting operation data of a high-voltage switch cabinet, and preprocessing the data through a method of deleting extreme abnormal values and complementing missing values; on the basis of a field investigation result and expert experience, selecting characteristic quantities for evaluating the high-voltage switch cabinet as a partial discharge value, an ultrasonic value, infrared diagnosis, a cable joint temperature and a busbar temperature; then, combining a polynomial function and a Gaussian radial basis function to construct a mixed kernel function, and mapping an original input space to a high-dimensional feature space; then optimizing a kernel function proportionality coefficient, a kernel function width parameter and a penalty factor by using an improved fruit fly optimization algorithm; and finally, by training an OVO-SVM high-voltage switch cabinet state estimation classifier, realizing nonlinear classification of the high-voltage switch cabinet state, namely, high-voltage switch cabinet state evaluation. According to the invention, multi-source information is utilized to carry out state comprehensive evaluation on the distribution high-voltage switch cabinet, a theoretical framework is simple, and the method is easy to implement and high in generalization ability.

The invention discloses an oil and gas reservoir parameter interpretation method and device based on a neural network and electronic equipment. The method comprises the steps of receiving to-be-interpreted parameter and a corresponding numerical range; sampling the to-be-interpreted parameter in the numerical range to obtain a trial calculation example; respectively inputting the trial calculationexample into a first Gaussian radial basis function neural network and a second Gaussian radial basis function neural network to obtain corresponding calculation pressure change data and calculationpressure derivative data; and calculating a target function value, if the target function value is smaller than or equal to judgment threshold, taking the parameter value contained in the current trial example as an interpretation result of the to-be-interpreted parameter, and if the target function value is greater than the judgment threshold, executing the step of sampling in the numerical rangeand the subsequent steps again until the interpretation result of the to-be-interpreted parameter is obtained. Based on the scheme provided by the invention, the efficiency and precision of numericalwell test interpretation can be improved, and the working intensity of well test interpretation personnel is reduced.

The invention discloses a soft robot state feedback control method based on a dielectric elastomer actuator. The method comprises the steps of building a dynamic control model of the dielectric elastomer actuator in a mode of virtual work simulation; building a Gaussian radial basis function neural network approximator for an unknown system function in the dynamic control model; by correspondinglydesigning a parameter updating rate, dynamically adjusting the parameters of the Gaussian radial basis function neural network approximator on line to realize the on-line approximation of the unknownsystem function; and finally, building a state feedback intelligent controller embedded with the Gaussian radial basis function neural network approximator in a state feedback control mode for dynamically adjusting the control input of the dielectric elastomer actuator on line in order to enable a state tracking error to converge to zero to realize a control target of a soft robot. According to the method, the control precision and the response speed of a control system are enhanced, and the adaptability and the intelligence of the soft robot based on the dielectric elastomer actuator are improved.

The embodiment of the invention provides a mechanical arm point-to-point trajectory planning method based on a radial basis function. The method comprises the steps of constructing a motion trajectory of a mechanical arm by adopting a Gaussian radial basis function; and constructing a corresponding optimization target equation according to parameters of the radial basis function, and optimizing the time of the whole trajectory. According to the method of the invention, the smooth motion trajectory can be generated between any target points, and any order derivative of the trajectory is continuous and is not influenced by the number of the target points. Meanwhile, the generated trajectory is simple in form and small in calculated amount, important parameters such as acceleration and jerk of the trajectory can be optimized through unified parameters, the smoothness of the trajectory is improved, then the running time of the whole trajectory is shortened, and the smoothness and working efficiency of the mechanical arm are improved.