17 results about "Polynomial kernel" patented technology

The embodiment of the invention discloses an insulator pollution flashover dynamic monitoring and early warning system, and relates to a power equipment state monitoring technology, and the system comprises a data collection module which is used for collecting the multi-source monitoring data of the operation state of an insulator, and the multi-source monitoring data comprises an electrical parameter, an environmental parameter and a pollution parameter; the feature processing module is used for performing time-frequency analysis and coupling analysis on the multi-source monitoring data to obtain data after feature processing; the mixed kernel function support vector machine analysis module comprises a kernel function calculation engine, a dynamic weight adjustment unit and a model optimizer, the kernel function calculation engine is used for supporting parallel calculation of a Gaussian kernel function and a polynomial kernel function, the dynamic weight adjustment unit is used for adaptive adjustment based on feature importance, and the model optimizer is used for parameter optimization; the early warning output module is used for outputting pollution flashover early warning information and maintenance decision suggestions according to the analysis result; the problems that a common monitoring system is poor in adaptability to complex environments and low in early warning accuracy can be solved.

The embodiment of the invention discloses an insulator pollution flashover dynamic monitoring and early warning method, device and equipment and a storage medium, and relates to the technical field of power equipment state monitoring, and the method comprises the steps: collecting the multi-source monitoring data of the operation state of an insulator in real time, the multi-source monitoring data comprising an electrical parameter, an environmental parameter and a pollution parameter; performing feature fusion processing on the multi-source monitoring data to construct a feature matrix, and obtaining an environment-pollution dynamic coupling factor; a hybrid kernel function support vector machine model is constructed, the hybrid kernel function support vector machine model adopts a hybrid kernel function of a Gaussian kernel and a polynomial kernel, and the kernel function proportion is dynamically optimized through an adaptive weight adjustment mechanism; based on the feature matrix and the environment-pollution dynamic coupling factor, using the mixed kernel function support vector machine model to realize pollution flashover early warning; the problems that a traditional monitoring method is poor in adaptability to a complex environment and low in early warning accuracy can be solved.

The invention discloses a laying hen lossless data processing method based on improved XGBoost, and relates to the technical field of data processing, discrete feature numeralization is realized through adaptive tag coding, abnormal value detection and non-linear mapping correction are combined, a robustness purification feature set is constructed, and a layer lossless data processing method based on the improved XGBoost is obtained. The problems that a traditional method is sensitive to sensor noisy points and lacks an effective data cleaning mechanism are solved, and prediction precision reduction caused by extreme sample interference is avoided. And performing nonlinear expansion on the purification features by using a polynomial kernel, and accurately capturing a coupling relationship between the features. The global optimal hyper-parameter optimization of the XGBoost integrated regression model is driven by high-dimensional features, so that the parameter optimization efficiency is improved, local optimum is avoided, the method adapts to a multi-source heterogeneous and high-noise nonlinear data scene of laying hen breeding, the processing precision and generalization ability of the model are remarkably improved, reliable data support is provided for accurate management and quality monitoring of laying hen breeding, and the method is suitable for large-scale popularization and application. And efficient landing of the intelligent breeding technology is promoted.

The present application relates to a kind of short-term load prediction method based on ISSA-HKELM, the method includes: first, in view of the defect of kernel extreme learning machine KELM, combined with Gaussian kernel function and polynomial kernel function, construct the hybrid kernel extreme learning machine HKELM with stronger generalization ability;Second, in view of the problem that sparrow search algorithm is easy to fall into local extremum, adaptive t distribution strategy and dynamic adaptive weight are introduced to improve sparrow search algorithm;Third, the improved sparrow search algorithm ISSA is used to optimize the parameter of hybrid kernel extreme learning machine HKELM and establish ISSA-HKELM prediction model;Finally, short-term load prediction is carried out using the established ISSA-HKELM model.Compared with prior art, the present application has good prediction accuracy and robustness and the like advantages.

The present application belongs to the field of building, especially a kind of rigid-flexible combination bagged pile vertical bearing capacity calculation method and system, the method specifically includes: obtaining the initial parameter set that influences pile vertical bearing capacity, including pile geometry, bag performance, soil mechanical parameters, and the one-dimensional control point vector of bottom expansion form extracted based on point cloud data reconstruction by non-uniform rational B spline, divided into side resistance parameter subset and end resistance parameter subset;Using historical sample training set, construct the hybrid kernel matrix of polynomial kernel and radial basis kernel combination for two subsets respectively, and weight according to the sensitivity difference of parameter to resistance;Mean centering and feature decomposition are carried out to the hybrid kernel matrix, and the principal component is selected according to the cumulative variance contribution rate and parameter correlation coefficient absolute value and threshold value rule, and the core factor of side resistance and end resistance is obtained by decoupling;Gaussian process regression model is established respectively with the two core factors, the side friction resistance and end resistance are predicted, and the sum is obtained to obtain the vertical bearing capacity of pile.

The invention relates to the technical field of fault detection, and discloses a transformer substation fault detection method and system, and the method comprises the steps: carrying out the windowing processing of collected time series data of a transformer substation, extracting a first feature vector, and calculating a second feature vector; constructing a mixed kernel function formed by weighted combination of a radial basis kernel function and a polynomial kernel function, and calculating weight coefficients of the radial basis kernel function and the polynomial kernel function; based on a preset model parameter and a mixed kernel function, training a single-class support vector machine by using the training sample under a normal working condition to obtain a fault decision boundary; calculating an early warning threshold, and setting an early warning boundary in the fault decision boundary; and performing feature extraction which is the same as that of the training sample on to-be-detected data, and inputting the to-be-detected data into the trained single-class support vector machine to judge the state of the to-be-detected data. According to the invention, the accuracy of fault detection is improved, the early symptom of equipment state deterioration can be identified, and early warning information is provided for operation and maintenance personnel.

The invention provides a near-field electromagnetic wave image reconstruction method based on information retention type mixed regularization sparse kernel learning, and belongs to the technical field of near-field electromagnetic wave image reconstruction. The method comprises the following steps: S1, carrying out data acquisition and preprocessing, carrying out normalization processing, and eliminating low-frequency drift; s2, mapping the input data to a feature space by using a Gaussian kernel or a polynomial kernel, and designing a dynamic dictionary updating strategy; s3, carrying out online recursive weight updating; s4, hybrid regularization reconstruction optimization is carried out, and a vector to be optimized is solved through fitting constraint conditions; s5, using FISTA to accelerate neighbor projection and a dual decomposition framework to iteratively solve an optimization problem, and estimating a step length and a momentum item through a spectral norm; s6, sparsification and dictionary maintenance are carried out, and a covariance matrix and a weight coefficient are updated; through the combination of mixed regularization and online kernel learning, the defects of a traditional CS method and a KRLS variant are overcome, real-time online updating, sparse prior and edge preservation are integrated, and the defects of a traditional near-field electromagnetic wave image reconstruction technology are overcome.

The application discloses an apple edge early damage detection method based on point cloud and hyperspectral imaging, and belongs to the field of apple edge early damage detection, and comprises the following steps: S1, synchronously collecting hyperspectral images and three-dimensional point cloud data of to-be-detected apples; S2, based on preprocessed multi-source heterogeneous data, realizing pixel-level alignment of the hyperspectral images and the three-dimensional point cloud data, and calculating surface normal vectors and effective incident angles; S3, based on a Lambert reflection model of a polynomial kernel function with a variable index, performing adaptive spectral correction; and S4, using a random forest algorithm to classify the corrected spectral data. The apple edge early damage detection method based on point cloud and hyperspectral imaging has the advantages that through fusion of three-dimensional point cloud and hyperspectral imaging data, combined with spatial registration, angle correction and a random forest algorithm, multi-source heterogeneous data are collaboratively analyzed, and the precision and reliability of apple edge early damage detection are significantly improved.

The application discloses a genomic selection method for machine learning of additive and non-additive effects, and by adding non-additive effects to a GBLUP model, the prediction accuracy of the GBLUP model is greatly improved. By applying a kernel function in the GBLUP, the model can better simulate the complex genetic relationship between genes and the nonlinear combination of gene effects, thereby improving the explanation ability of genetic variation. In addition, a machine learning strategy K P RR is constructed based on a polynomial kernel function and a kernel ridge regression. By combining the advantages of the polynomial kernel function and the kernel ridge regression method, the K P RR strategy allows remote data points to contribute to the value of the model, which enables the interaction in the genomic data to be more effectively contributed, and the genomic prediction effect can be greatly improved.

This invention discloses a target recognition method based on adaptive multinomial kernel function and multimodal fusion, belonging to the field of target recognition technology. The method includes: acquiring the original signal of the target to be identified; preprocessing the original signal using a wavelet threshold denoising algorithm to generate a signal to be analyzed; performing multimodal time-frequency analysis on the signal to be analyzed, outputting three types of time-frequency feature maps; dynamically weighting and fusing the three types of time-frequency feature maps to generate a fused time-frequency feature tensor, and then processing it through a convolutional neural network to generate a time-series feature vector; inputting the time-series feature vector into a convolutional neural network and a gradient boosting regression tree (GBRT) model respectively for processing, and then outputting preliminary recognition results and feature verification results based on the target category to complete target recognition. This invention achieves accurate target recognition by combining the advantages of multimodal analysis technology with short-time Fourier transform and wavelet transform.

The embodiment of the invention discloses a hybrid kernel function support vector machine model construction method, and relates to the technical field of power equipment state monitoring, and the method comprises the steps: constructing a hybrid kernel function support vector machine model, the model adopts a mixed kernel function of a Gaussian kernel function and a polynomial kernel function, the Gaussian kernel function is used for capturing local features of data, and the polynomial kernel function is used for capturing global features of the data; dynamically optimizing a kernel function proportion through an adaptive weight adjustment mechanism; training the mixed kernel function support vector machine model by using a historical data set, and optimizing model parameters including penalty factors and kernel function parameters: processing the collected multi-source monitoring data of the marginal operation state based on the trained mixed kernel function support vector machine model to realize pollution flashover early warning; a model suitable for classification and prediction of a complex data set can be constructed, such as dynamic monitoring and early warning of pollution flashover of an insulator of a power system, so as to improve prediction accuracy and adaptability.

This invention relates to the field of fault detection technology and discloses a substation fault detection method and system. The method includes: windowing the collected substation time-series data to extract a first feature vector and calculate a second feature vector; constructing a hybrid kernel function composed of a weighted combination of radial basis function (RBF) kernel functions and polynomial kernel functions, and calculating the weight coefficients of the RBF kernel function and polynomial kernel function; training a single-class support vector machine (SVM) using training samples under normal operating conditions based on preset model parameters and the hybrid kernel function to obtain the fault decision boundary; calculating an early warning threshold and setting an early warning boundary within the fault decision boundary; and performing the same feature extraction on the test data as on the training samples and inputting it into the trained SVM to determine the state of the test data. This invention improves the accuracy of fault detection, can identify early signs of equipment deterioration, and provides early warning information for operation and maintenance personnel.

The invention discloses a polynomial kernel-based support vector regression model reverse inversion simulation parameter optimal value method, and relates to the technical field of numerical simulation parameter calibration. According to the method, a data set is constructed by designing a simulation parameter value combination, a prediction model is constructed by adopting polynomial kernel function support vector regression (Poly-SVR), hyper-parameter optimization is realized by combining 5-fold cross validation and grid search, and finally, reverse mapping of a response quantity and a simulation parameter is completed by utilizing an L-BFGS-B algorithm. The method solves the problems that a traditional response surface method is low in calibration precision and high in overfitting risk in a small sample scene, achieves efficient and accurate inversion of simulation parameters, and is suitable for various engineering application scenes where the simulation parameters need to be reversely determined through the actual measurement response quantity.

The invention provides a single-feature brain fatigue recognition method based on multi-kernel learning. The method comprises the steps that firstly, L different kernel functions are subjected to weighted combination to form a combined kernel function of an MK-SVM; secondly, optimizing the weight of a combined kernel function through kernel matrix combination and by using a kernel function standardization and cross validation method, and forming an improved MK-SVM multi-classification model; and finally, the electroencephalogram signals are classified by the MK-SVM multi-classification model through weighted combination of a plurality of kernel functions, behaviors of the kernel functions are adjusted according to specific tasks, and an optimal decision boundary is searched in different kernel spaces. According to the multi-kernel SVM, a plurality of basic kernel functions such as a linear kernel, a Gaussian kernel and a polynomial kernel are combined to form a composite kernel, the weight of each kernel is optimized, and the flexibility and robustness of the model are improved so as to adapt to electroencephalogram feature complex data distribution and improve electroencephalogram signal classification indexes.

The invention discloses a network flow detection method and device based on a quantum support vector machine, and relates to the field of network data security, and the method comprises the steps: carrying out the quantum feature coding of each flow in to-be-detected network flow data, and obtaining a quantum coding representation; wherein the quantum feature coding is carried out by mixing amplitude coding and angle coding; based on a quantum support vector machine model, calculating the quantum kernel similarity between the quantum coding representation of each flow and each support vector in the quantum support vector machine model, and according to the quantum kernel similarity of each support vector, calculating a flow category dichotomy value of the quantum coding representation of each flow; wherein the quantum support vector machine model is obtained by training based on a quantum kernel combining a Gaussian kernel and a polynomial kernel; and obtaining a detection result of each piece of flow according to the flow category dichotomy value of each piece of flow. According to the embodiment of the invention, the technical problem that the network flow detection precision and efficiency of the existing energy storage system are not expected can be solved.

The invention relates to a load prediction multidimensional data preprocessing method. The method comprises the following steps: firstly, acquiring a load data time sequence formed by multi-dimensional load data; then decomposing the load data time sequence to obtain a trend component which refers to a long-term trend sequence in the load data time sequence; obtaining an inflection point of the load data time sequence according to the trend component; approximation is carried out on the load data at the non-inflection point by using a linear kernel function, approximation is carried out on the load data at the inflection point by using a preset adaptive polynomial kernel function, and then a first kernel function is obtained according to the linear kernel function and the adaptive polynomial kernel function; constructing a kernel matrix according to the first kernel function; and further performing non-linear analysis on the kernel matrix by using a factor analysis method to obtain load data after dimension reduction, and finally achieving the purposes of simplifying observation data and explaining and researching complex problems by using a small number of variables.

The invention discloses a target recognition method based on an adaptive polynomial kernel function and multi-modal fusion, and belongs to the technical field of target recognition, and the target recognition method comprises the steps: obtaining an original signal of a to-be-recognized target, and carrying out the preprocessing of the original signal based on a wavelet threshold denoising algorithm, and generating a to-be-analyzed signal; performing multi-modal time-frequency analysis on the to-be-analyzed signal, and outputting three types of time-frequency characteristic patterns; carrying out dynamic weighted fusion processing on the three types of time-frequency feature maps to generate a fused time-frequency feature tensor, and carrying out convolutional neural network processing to generate a time sequence feature vector; and respectively inputting the time sequence feature vector into a convolutional neural network and a gradient boosting regression tree GBRT model for processing, and then outputting a preliminary recognition result and a feature verification result based on a target category to complete target recognition. According to the method, through a multi-modal analysis technology, the advantages of short-time Fourier transform and wavelet transform are combined, and accurate recognition of the target is achieved.