33 results about "Gram schmidt" patented technology

The present invention relates to a super-directivity beamforming method based on modal decomposition and synthesis. Firstly, the optimal solution of the beamforming theory is decomposed and expressed as a matrix; then the correlation matrix is ​​given by using Gram-Schmidt orthogonal transformation Then, the optimal beam is decomposed into modal beams with different directivity and robustness by using the matrix form of the optimal solution, and the maximum directivity factor is obtained by superimposing the directivity factors of each modal beam. Finally, taking the uniform linear array as an example, the final robust super-directional beam is obtained by selecting the appropriate modal beam synthesis. The present invention utilizes the following properties to overcome the deficiencies of inaccuracy and limited application range of the prior art: the optimal solution of the beamforming theory is not limited by the array form, and the Gram-Schmidt orthogonal transformation has an accurate recursive calculation formula.

The invention proposes a common vector based sparse representation classification method. The method comprises the steps of firstly, performing Gram-Schmidt orthogonal transformation through a differential sub-space of each training sample type to obtain a common vector of each type; secondly, by taking a dictionary composed of all common vectors as a dictionary of a sparse representation classifier, calculating a sparse coefficient of a test sample in the dictionary through l1-norm minimization; and obtaining an estimated test sample by using the sparse coefficient corresponding to each type and the training sample of the type, comparing the estimated test sample with the acquired test sample, and taking the type with the highest similarity as a classification result. The core thought of the invention is that a dictionary, consisting of all training samples, of an original sparse representation classifier is replaced with the dictionary consisting of the common vectors of all types in the training samples, thereby remarkably increasing the correctness rate of a human face identification method in a small sample condition.

This application discloses a method and device for realizing QR decomposition of a matrix. The method includes: obtaining a complex received signal vector yc and a complex channel matrix Hc; combining and transforming the complex received signal vector yc and the complex channel matrix Hc to obtain a real matrix C; By using the coordinate rotation digital calculation algorithm, the elements in each column of the real matrix C are eliminated to obtain the upper triangular matrix R and the vector QTy at the same time. In the method disclosed in the embodiment of the present application, the obtained complex received signal vector and the complex channel matrix are combined and transformed to obtain a specific real matrix, and then the elements in each column of the above real matrix are eliminated by using the Cordic algorithm , obtain the upper triangular matrix R, and obtain the vector QTy at the same time, it can be seen that, with respect to the Gram-Schmidt method, the present invention saves the hardware resources occupied when additionally calculating QTy, that is, reduces the cost of obtaining the upper triangular matrix R and QTy The hardware resources occupied by the process.

The invention relates to a gear fault diagnosis method based on the orthogonal match between multiple parallel dictionaries. According to the method, gear vibration signals are expressed in the mode of linear superposition of simple and sparse atoms of the multiple parallel dictionaries. As for the multiple parallel dictionaries, Fourier dictionaries and impact time frequency dictionaries are selected to form the multiple dictionaries according to the characteristics of the gear vibration signals, matched atoms are selected in parallel in all sub-dictionaries with a genetic algorithm, coefficients of all orders are compared to obtain a most matched atom, Gram-Schmidt orthogonalization is performed on the atom, and then a new atom library is formed. Analysis signals are projected to the atom library, and the projections are subtracted from the signals to form residual signals to be decomposed the next time. The decomposition process is completed after the iteration end conditions are met, the matched atoms and the matching coefficient are extracted, the matched atoms based on the impact time frequency dictionaries are reconstructed, corresponding impact components can be obtained, and then fault information of the gear vibration signals is demodulated and extracted for fault diagnosis.