37 results about "Nonlinear manifold" patented technology

The invention relates to a surge monitoring method based on incremental nonlinear manifold learning, high-dimensional characteristic information is constructed according to multipath dynamic characteristics expressed in the operating process of a compressor, a one-dimensional main manifold is extracted by utilizing a local tangent space algorithm, the one-dimensional main manifold is updated in real time by an incremental manifold learning method, and whether a surge occurs or not under the varying duty condition of the compressor is judged by monitoring the change of a geometrical structure of a main manifold time sequence. The monitoring method extracts a concealed nonlinear change rule from complete high-dimensional characteristic data in the self operation of compressor equipment, thereby avoiding a phenomenon of missing report; meanwhile, the main manifold has the characteristic of enlarging diversity, and on the basis of ensuring real-time performance, an incremental algorithm simplifies the setting of a warning line and achieves low false alarm rate, and lays a foundation for effectively realizing surge early warning.

The invention discloses an image super-resolution reconstruction method based on maximum linear block neighborhood embedding, the main steps of which include: constructing a training sample set, clustering by using a hierarchical splitting clustering method, and approximately dividing the nonlinear manifold into multiple The largest linear block, after clustering, uses medium and high frequency features to construct the largest linear block; classifies low-resolution test images into edge blocks and non-edge blocks, and uses two different neighborhood selection methods to reconstruct the results more accurately ; Neighborhood selection; Neighborhood embedding; Image reconstruction, deblurring the initial reconstructed image to obtain a complete and clear high-resolution reconstructed image. The present invention approximates a plurality of maximum linear block structures from the nonlinear manifold of training samples through a clustering method, combines feature representation and neighborhood selection to realize local linear neighborhood embedding, reconstructs more accurate high-frequency information, and greatly reduces The time complexity is reduced, and the super-resolution reconstruction of natural images can restore clearer edge details.

The invention relates to an electromechanical device nonlinear failure prediction method, comprising the following steps: 1, obtain data which can represent the running state of a device and select a section continuous vibration signal which has a long course and is sensitive to the failure to analyze; 2, respectively carry out exceptional value elimination and missing data filling to the vibration data by a 3 sigma method and an interpolation method; 3, carry out noise reduction to the vibration signal by a lifting wavelet method; 4, decompose the vibration signal after the noise reduction to corresponding characteristic bandwidths; 5, obtain a low dimension manifold character by utilizing a typical predicted characteristic bandwidth and adopting a nonlinear manifold learning method through decoupling of topological mapping and non-failure energy information; 6, carry out intelligent failure prediction with long course trend in a time domain by utilizing a recurrent neural network which has the dynamic self-adaptive characteristic and a first dimension of the low dimension manifold character as a neural network input. The lifting wavelet method is adopted in the invention, the algorithm is simple, the arithmetic speed is high, and the used memory is less, thereby being suitable for the characteristic bandwidth abstraction of failure character. The electromechanical device nonlinear failure prediction method can be widely applied to the failure prediction of all kinds of electromechanical devices.

The present invention discloses a method for synthesizing three-dimensional human body movement based on non-linearity manifold study, so as to make three-dimensional human body movement animations, characterized in that firstly a set of sparse three-dimensional human body movement samples is mapped in movement semantic parameter space builded on a low-dimentsion manifold; then implementing uniformly distributed coarctation resample to the low dimensional movement semantic parameter space, and applying resample coefficient set to movement samples distributed in an original movement space sparsely to obtain dense and well distributed movement samples of a high dimensional space; then remapping the newly sampled high dimensional movement samples to obtain a final low dimensional movement semantic parameter space; finally, by means of interacting the movement semantic parameters synthezed selectively in the low dimensional semantic parameter space by users, the system maps the movement semantic parameter to a high dimensional movement space to obtain a new movement sequence. The invention is not only capable of controlling precisively movement physical parameters, e.g. movement position, physical movement characteristics of special arthrosis, and also used to synthesize novel movement data having high-rise movement semantion such as movement styles.

The invention provides a nuclear power plant fault diagnosis method based on local linear embedding and K-nearest neighbor classifier. (1) Obtain the operating data of nuclear power plants in steady state operation and typical accident conditions as training data; (2) Use the mean-variance standardization method to perform dimensionless standardization on the training data to obtain high-dimensional sample data; (3) Use the local linear embedding algorithm to extract the low-dimensional manifold structure of high-dimensional sample data, and obtain the low-dimensional feature vector; (4) input the low-dimensional feature vector into the K-nearest neighbor classifier for classification training; (5) obtain the nuclear power plant Run the data in real time, repeat (2), (3); (6) use the trained K-nearest neighbor classifier to classify the feature vectors. The invention utilizes the advantages of the nonlinear manifold learning method in feature dimension reduction and extraction, is suitable for fault diagnosis of nonlinear and high-dimensional data systems, and has high fault diagnosis accuracy.

The invention discloses a vehicle intrusion detection method and device. The steps of the method include: collecting high-dimensional CAN data on the vehicle-mounted CAN bus during vehicle operation; Clustering; using the nonlinear manifold learning method to obtain the low-dimensional manifold of the detection data of each data cluster; collecting CAN data in a simulated normal driving environment, constructing a normal data set and training to obtain a standard low-dimensional manifold; according to the characteristics of the vehicle, the detection The data low-dimensional manifold is compared with the standard low-dimensional manifold in three-dimensional space to determine whether the vehicle has an intrusion event. The invention performs intrusion detection based on the geometry of CAN data, has fast detection speed, satisfies the continuous delivery of CAN messages during the driving process of the vehicle, and can detect the vehicle in real time.