31results about How to "Solving the Curse of Dimensionality" patented technology

The invention discloses a method for constructing a five-freedom bearingless asynchronous motor support vector machine inverse controller. A support vector machine inverse with 7 input nodes and 7 output nodes is formed by using a support vector machine and 12 integrators, the support vector machine having 19 input nodes and 7 output nodes. The support vector machine inverse is disposed in front of a composite controlled object to form a pseudo-linear system. The pseudo-linear system is equivalent to five positional second-order integral pseudo-linear subsystems, a rotate speed first-order integral pseudo-linear subsystem, and a flux linkage first-order integral pseudo-linear subsystem. Corresponding controllers are respectively designed, and a linear closed loop controller is formed. A support vector machine inverse controller is formed by the linear closed loop controller, the support vector machine inverse, three Clark inverse transformation, three current-tracking inverters, and a linear power amplifier, thereby realizing non-linear dynamic decoupling control among suspension force, rotating speed, and flux linkage.

The invention discloses a multi-feature locality sensitive hashing (LSH) indexing combination-based remote sensing image retrieval method and belongs to the technical field of remote sensing image retrieval. According to the multi-feature LSH indexing combination-based remote sensing image retrieval method disclosed by the invention, LSH indexing of one of the best indexing technologies in high-dimensional feature spaces is introduced into the field of the remote sensing image retrieval, so that the problems of curse of dimensionality and retrieval time consuming can be effectively solved on a large scale, and the rapid retrieval of remote sensing images is realized. Meanwhile, the invention provides a new indexing validation index-a feature discriminative-ness-based indexing validation index (FDIVI) by aiming at the LSH indexing, and features best capable of distinguishing targets and backgrounds are evaluated and selected by the LSH indexing in all feature spaces, and therefore, the accuracy of a retrieval result is effectively improved. Compared with the prior art, the multi-feature LSH indexing combination-based remote sensing image retrieval method disclosed by the invention is capable of more rapidly and accurately realizing the retrieval of a great amount of remote sensing image data.

The invention relates to a wireless self-backhaul small base station access control and resource allocation joint optimization method, which belongs to the field of wireless communication. The method includes: under the premise of the average delay constraint of each user and the transmission power constraint of each base station, using the Markov decision process to establish a joint maximum spectrum efficiency and wireless resource allocation method for wireless self-backhaul small base stations A multi-objective optimization model for energy efficiency; in the environment of random dynamic arrival of user data packets, an approximate dynamic programming method is used to formulate the best access and resource allocation strategy for any dynamically arriving user demand, so that the system can operate in a relatively long-term Higher spectrum efficiency and energy efficiency can be obtained in resource allocation. The wireless self-backhaul small base station access control and resource allocation joint optimization method proposed by the present invention can maximize the long-term average spectrum efficiency and energy efficiency while ensuring the average delay constraint and transmission power constraint.

The present invention proposes an adaptive container migration method in a fog computing environment, which includes the following steps: establishing a container-based fog computing framework, the container is located on the fog node, the mobile application is located on the user, and the user's task is executed in the container; Modeling the target of container migration in the fog computing scenario, the migration target includes delay, power consumption and migration overhead; setting the state space and action space, defining the reward function and setting the Q iteration function; through the deep neural network Dimensionality reduction of the state space; through the optimization of action selection, the dimensionality reduction of the action space is realized. Finally, a prototype of a container adaptive migration system was implemented and the entire process was verified. The adaptive container migration method in the fog computing environment proposed by the present invention can better plan the resources in the fog computing, reduce the time delay between the user and the fog nodes, and reduce the energy consumption of the fog nodes.

The present invention provides a coordinated optimization method, system, equipment, and storage medium for an electric-thermal integrated energy system. The method includes: obtaining real-time electric-thermal integrated energy system parameters; The power generation power of the power system, thermal system and coupling device of the thermal integrated energy system; input the generated power into the pre-trained optimal scheduling model based on the SAC framework, output scheduling actions, and form an electric-thermal integrated energy system coordination strategy. Scheduling actions can be given directly through the trained policy network, without the need for traditional nonlinear overall iterative solutions, the calculation speed has been significantly improved, and the calculation efficiency is higher.

The invention discloses a hyperspectral image classification method based on a cross-grouping space-spectral feature enhancement network, comprising: 1) spectral feature cross-grouping; 2) multi-channel grouping spectral feature extraction; 3) spatial feature cross-grouping; 4) grouping space Feature extraction; 5) Spectral-spatial channel information interaction; 6) Hyperspectral image pixel classification. This method uses spectral-spatial feature information to perform cross-grouping and feature extraction operations on spectral features and spatial features, which can effectively reduce the correlation between adjacent spectra. The obtained features are enhanced, information interaction and fusion of spatial features and spectral features are performed, and the fused features are used for classification, which can improve the network classification performance.

The invention discloses an indoor scene recognition method combining deep learning and sparse representation. Discriminate and detect object categories with test samples to construct the bottom-level features of each indoor scene image; use the bag-of-words model to combine the bottom-level features and spatial features of each indoor scene image to construct mid-level features; The middle-level features are combined to construct a sparse dictionary; the sparse dictionary is used to sparsely represent the test samples, and the residual is calculated according to the obtained sparse solution and the input test sample, and the object category of the test sample is judged according to the size of the residual ; It will be judged to get the output of the object category it belongs to. The invention can accurately identify indoor scenes, can effectively improve the accuracy and robustness of indoor scene identification, and has high practical performance.