9 results about "Self-organizing map" patented technology

A voice generation method and apparatus, a product, a device, and a medium, which are applied to the technical field of voice generation. The method comprises: using a quantizer to discretize a voice feature vector of an original voice signal, obtaining a discrete symbol representation corresponding to the original voice signal (S11); extracting a text feature corresponding to a target text (S12); and inputting the text feature and the discrete symbol representation into a voice generation model, so that the voice generation model uses the text feature as a condition, and generates a target voice on the basis of the discrete symbol representation (S13). The quantizer is obtained by training a self-organizing map network by means of voice signal training samples. The method may restore an original voice feature more accurately, and improve the quality of the generated voice.

A system and method for training relational networks. A method includes applying a self-organizing map (SOM) to training data in order to create a visualization. The SOM is a neural network configured to transform relationships between data items. The visualization has a lower dimensionality than the training data. The method also includes training machine learning models of a generative relational network (GRN) based on the visualization, where the GRN includes sets of nodes having respective machine learning models among the machine learning models of the GRN and the sets of nodes include a set of dominance factor nodes and a set of evolution of internal component nodes. The set of dominance factor nodes defines a dominance factor based on change intensity and change frequency, and the set of evolution of internal component nodes defines evolution with respect to changes determined based on values of the dominance factor over time.

This paper presents a method and system for correcting and predicting supersonic internal flow fields by integrating topology consistency evaluation and physical constraint latent space mapping, belonging to the interdisciplinary fields of fluid aerodynamics design and artificial intelligence. The method first employs a topology consistency evaluation based on a self-organizing mapping grid to realize the nonlinear coupling evolution of parameter sets and spatial errors under supersonic conditions in a hexagonal topological space. Gaussian smoothing and dot product operations are used to quantify the topology consistency between parameters and spatial errors, automatically selecting core parameters. Next, a physical constraint latent space mapping architecture is constructed, introducing a composite physical loss function to reduce the dimensionality of high-dimensional flow field features and establish a mapping model from core parameters to the latent space. Decoder weights are frozen to ensure the continuity of physical laws and derivatives. Finally, an error-driven correction mechanism is used to statistically analyze residuals and generate an error feedback matrix to complete the implicit core parameters, achieving closed-loop reconstruction and corrective prediction of the surrogate model. This method can significantly improve the physical fidelity of supersonic internal flow field reconstruction and effectively solve the problem of large prediction deviations in the flow field behind the gate.

This invention discloses a reservoir prediction method and system based on multi-source information fusion, belonging to the field of reservoir prediction technology. The method includes: high-fidelity signal processing of seismic data combined with seismic geological conditions of the target area; fine well-seismic calibration using well logging, seismic, and geological information combined with domain knowledge; seismic facies identification using a deep embedded self-organizing map network; multi-scale fault detection using cepstral analysis-enhanced coherent ant body attributes; wave impedance and porosity inversion using seismic waveform indication inversion; reservoir gas-bearing capacity detection by combining data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data; and prediction of favorable reservoir distribution by adaptively weighted fusion of the results of seismic facies identification, fault detection, wave impedance and porosity inversion, and gas-bearing capacity detection. The advantage of this invention is that it can achieve accurate and reliable prediction of deep and complex oil and gas reservoirs by integrating multi-source, multi-dimensional, and multi-attribute data.

The present application relates to the technical field of power system operation optimization, and discloses a wind-solar-hydro-storage collaborative operation optimization method based on weather typing and volatility characteristics, comprising: collecting historical output data and meteorological data of wind power, photovoltaic, hydroelectric and energy storage systems, and constructing meteorological feature vectors and volatility feature vectors; inputting the comprehensive feature vectors fusing the meteorological feature vectors and the volatility feature vectors into a self-organizing mapping network for clustering; for each weather scenario, calculating the linear correlation, nonlinear mutual information and timing similarity among wind, solar, hydro and storage, constructing a collaborative relationship matrix, and extracting the average synergy degree, volatility offset rate and energy dominant coefficient as the scenario characteristics; establishing a multi-energy capacity optimization model, and solving the optimal capacity configuration by using a particle swarm optimization algorithm. The present application has the advantages of significantly improving the multi-energy complementary utilization efficiency in a high-proportion new energy system, reducing the energy storage configuration cost, and enhancing the stability and reliability of system operation.

This invention relates to the field of ecological environment monitoring and carbon cycle assessment technology, and discloses a forest carbon sink prediction method based on spatial heterogeneity decoupling and hierarchical distributed autoregressive meta-learning fusion. The method first integrates ground survey data, remote sensing temporal features, meteorological driving factors, topographic factors, and soil attribute data to construct a multi-source heterogeneous feature system. Then, it decouples and models different categories of features, and divides ecologically homogeneous clusters based on a self-organizing map neural network. For each ecological cluster, it constructs an independent nonlinear autoregressive model for distributed prediction. Based on this, a meta-learning fusion model is constructed, and the outputs of each distributed model are dynamically weighted and biased to obtain the predicted forest vegetation carbon density, and the forest vegetation carbon storage and annual carbon sink are calculated. This method can effectively handle spatial heterogeneity and temporal lag effects, constructs a progressive error convergence structure, and improves the accuracy, stability, and cross-regional generalization ability of forest carbon sink prediction.

The present application relates to the technical field of petroleum chemical production, more particularly to a crude oil property prediction method and system based on a self-organizing mapping network. The method comprises: collecting a batch of crude oil samples as reference benchmark crude oil samples, and dividing the obtained physicochemical property values into easy-to-measure vectors and difficult-to-measure vectors; collecting a crude oil sample to be measured, obtaining the corresponding easy-to-measure vectors, inputting the easy-to-measure vector data of the crude oil sample to be measured and the reference benchmark crude oil samples into a self-organizing mapping network for clustering, and obtaining a plurality of groups of crude oil vector clusters; screening reference benchmark crude oil samples under the same cluster as the crude oil sample to be measured; calculating the similarity of the screened reference benchmark crude oil samples and the crude oil sample to be measured, assigning a weight value to the difficult-to-measure vectors of the reference benchmark crude oil samples; calculating the difficult-to-measure vectors of the crude oil sample to be measured, and realizing the prediction of the crude oil properties of the crude oil sample to be measured. The present application provides reliable support for the rapid analysis of the overall physicochemical properties of crude oil.

The application discloses a reservoir group flood resource utilization risk source identification and scene set generation method. It comprises the following steps: constructing a risk condition hydrology-risk joint manifold, taking the target risk level as the front input condition, using a parameter mapping model to automatically generate a flood process, and through simulation deviation feedback closed loop iteration correction manifold parameters to generate a sample set with controllable risk distribution; inputting the sample set into a risk topology self-organizing mapping network, clustering based on hydrology-risk joint distance measurement and risk sensitive neighborhood rules, and constructing a hierarchical scene tree with risk level semantics; based on the scene tree, multi-stage risk budget and discount parameter optimization under the fairness constraint are carried out to generate a risk control scene set and identify the risk source. The application solves the problems of uncontrollable generation process, easy loss of high risk scene and uneven stage risk allocation in the traditional method, and realizes efficient scene generation and scientific scheduling in a risk-oriented manner.

This invention discloses a two-stage spatial clustering method for wind and solar resources based on VAESOM-Kmeans, belonging to the field of optimal scheduling of new energy power systems. This method acquires wind and solar resource feature data containing static intensity and dynamic fluctuation characteristics and constructs a normalized feature matrix; it uses a variational autoencoder (VAE) to extract low-dimensional continuous latent features from high-dimensional data to achieve dimensionality reduction and noise reduction; the latent features are input into a self-organizing map (SOM) neural network, and the optimal number of clusters and initial cluster centers are adaptively determined based on topological distribution; using these as parameters, the Kmeans algorithm is used to complete spatial clustering; finally, the main stations of each cluster interval are selected based on the Pearson correlation coefficient, and typical output sequences are output. This invention overcomes the limitations of traditional geospatial clustering and the defect of Kmeans blindly specifying the K value, significantly simplifying the node size of large-scale multi-energy complementary optimal scheduling models and improving solution efficiency.