5 results about "Random subspace method" patented technology

The application provides a cut learning system and a bandwidth-aware neural subspace compression method, device and equipment and medium thereof, and relates to the technical field of cut learning. The bandwidth-aware neural subspace compression method comprises: obtaining to-be-compressed data. According to the to-be-compressed data, a tensor singular value spectrum is estimated by an adaptive rank selection module based on a random subspace method, and is trimmed in combination with an energy coverage threshold, a bandwidth budget and a rank upper limit to obtain a compression rank. According to the to-be-compressed data, the compression rank and an error feedback item of a previous round of iteration, left and right factor matrices are alternately updated by an alternating orthogonal approximation module, and subspaces in row and column directions are orthogonalized to obtain row and column subspace orthogonal bases and a low-rank approximation. According to the to-be-compressed data and the low-rank approximation result, an error feedback item of the current round of iteration is obtained. The row and column subspace orthogonal bases are compressed data suitable for transmission.

The invention provides a method for identifying modal parameters of a supporting structure of a wind generating set, which comprises the following steps of: acquiring acceleration data of the supporting structure through a vibration data acquisition system, and identifying the modal parameters of the supporting structure by adopting a modal parameter identification algorithm based on a random subspace method, a clustering method and a modal shape denoising method. The modal parameters such as the vibration frequency, the damping ratio and the modal shape of the supporting structure are accurately extracted, and the health state evaluation and early warning of the supporting structure and the correction of a prediction model are realized through the constructed prediction model among the environment parameters, the operation parameters and the modal parameters and by utilizing an artificial intelligence edge calculation box. Finally, the operation strategy of the wind generating set and the frequency of the damping tuning vibration attenuation system are adjusted in time through the operation control system, and safe operation of the supporting structure is guaranteed.

The application relates to the field of non-contact monitoring of overhead transmission lines, and particularly provides a method and device for identifying the galloping mode of a transmission line based on magnetic field induction. The method comprises the following steps: obtaining the initial magnetic induction intensity of a magnetic dipole in each direction in a space coordinate system when a target overhead conductor gallops; wherein the initial magnetic induction intensity is measured by two magnetic induction sensors above the target overhead conductor; inversely calculating the conductor movement by using a nonlinear optimization method according to the magnetic induction intensity in each direction; wherein the conductor movement comprises the vertical displacement and the torsion angle of the target overhead conductor; and identifying the target mode of the target overhead conductor galloping by using a random subspace method based on the vertical displacement and the torsion angle. The technical scheme provided by the application can improve the accuracy of the galloping mode identification of the transmission line to a certain extent.

The application discloses a hybrid wind turbine tower modal test method and system, comprising the following steps: S1, obtaining target tower design parameters; S2, forming an overall and local test point array; S3, applying excitation in a manner of combining environmental excitation and artificial excitation and collecting response signals; S4, synchronously collecting acceleration, excitation force and environmental parameters through a multi-channel collector; S5, after data preprocessing, respectively adopting a random subspace method and a frequency response function modal analysis method to extract modal parameters, and obtaining final modal parameters through fusion correction; and S6, result verification and application, comparing the test results with a finite element model, correcting the model and outputting structure evaluation and optimization suggestions. The local modal of a cross rod, an inclined rod and a connecting node is captured, a weak link of the structure is effectively identified, the loading mode of environmental excitation+artificial excitation cooperation is adopted, and a modal confidence criterion is introduced for verification, so that the modal frequency test error is reduced.