48 results about "Pre whitening" patented technology

A MIMO OFDM system includes a plurality of space-time encoders for encoding respective data blocks with independent space-time codes. The transformed data block signals are transmitted by a plurality of transmit antennas and received by a plurality of receive antennas. The received data is pre-whitened prior to maximum likelihood detection. In one embodiment, successive interference cancellation can be used to improve system performance. Channel parameter estimation can be enhanced by weighting the channel impulse response estimates based upon a deviation from average.

The invention discloses a height measurement method based on an array interpolation compression perception. The height measurement method mainly aims at solving a low elevation height measurement problem under a multipath environment, and especially under low signal to noise ratio and less snapshot environments. The method comprises the following steps of extracting a target signal from a radar echo; acquiring a spatial-domain sparse signal through cancellation and signal reconstruction; using a wave beam formation method to obtain a rough measurement target angle; according to the rough measurement angle, acquiring the spatial domain and dividing the spatial domain; using the array interpolation to acquire a virtual array; according to a matrix transformation relation, acquiring an interpolation transformation matrix and carrying out prewhitening processing on the interpolation transformation matrix; using a whitening interpolation transformation matrix and an observation matrix to acquire an observation signal; using a whitening interpolation transformation matrix and observation signal iteration operation to acquire a target signal estimation value; extracting a target angle from the target signal estimation value so as to acquire a target height. By using the method of the invention, sampling points of the target signal and computation intensity are obviously reduced; sidelobes of a signal power spectrum and a space spectrum are effectively reduced; the method can be used in target tracking.

The invention provides a Wiener filtering speech enhancement method for multi-taper spectrum estimation of a pickup. The Wiener filtering speech enhancement method is characterized by including: performing multi-taper spectrum estimation on a noisy speech, and removing noise items through a wavelet threshold value to obtain speech spectra which are approximately pure; comparing the speech spectra with a speech spectrum processed through Wiener filtering, and selecting corresponding spectra as finally enhanced speech spectra according to different distortion types. When the method is applied in the pickup, compared with existing mean square forecast error and pre-whitening subspace methods, the method is smoother in speech and closer to pure speech; in circumstances with low signal-to-noise ratio and complex noise, the pickup is higher in speech processing speed, higher in coding and decoding efficiency, better in denoising performance and higher in robustness. In addition, in environments with long distance, the pickup has good speech recognition effect.

The invention relates to an interpolation transformation and beam forming-based far-field coherent signal DOA estimation method. According to the method, firstly, an interpolation matrix is adopted to convert a non-uniform linear array covariance matrix into a covariance matrix of a virtual uniform linear array. Data on the covariance matrix of the virtual array are subjected to noise pre-whitening to obtain img file = 'DDA 00013602494400000000000011.T'wi = '67 'he = '71 '. After that, img file = 'DDA 00013602494400000000000011.T'wi = '43 'he = '70 ' is subjected to spatial smoothing treatment to resolve the phase coherence, so that a coherence-resolved data covariance matrix is obtained. The coherence-resolved data covariance matrix is processed through constructing a cost function, and then the estimated value of the DOA of a far-field coherent signal is obtained. According to the invention, on the premise that the precision is guaranteed, the operations of feature decomposition, frequency spectrum searching and the like which are complicated in operation, can be avoided. The method is low in calculation complexity, simple and effective. Meanwhile, based on the method, the application range is popularized from a uniform linear array to any linear array, and from a non-coherent signal source to a coherent signal source. Therefore, the method is wider in application range.

The invention relates to a drum ball mill level detection method, in particular to a drum ball mill level detection method based on a multi-sensor technique, and solves the problem of low detection precision in the existing drum ball mill level detection methods. The method comprises the following steps: (1) respectively acquiring noise signals s1, s2 and s3 at a drum of a drum ball mill, the position at which self-background noise is higher and the environment in which the drum ball mill is located; (2) performing conditioning, analog to digital conversion and pre-whitening processing on the noise signals s1, s2 and s3 to obtain signals w1, w2 and w3; (3) respectively performing blind source separation processing on the signals w1, w2 and w3 to obtain signals x1, x2 and x3; (4) screening information representing drum ball mill level information from the signals x1, x2 and x3, and performing discrete Fourier transform, module taking and normalization on the information to obtain an output signal z(n); and (5) computing the value of a drum ball mill level L according to the output signal z(n). The detection method is reasonable, can effectively reduce interference of environmental noise and the self-background noise, and can perform accurate and reliable detection on the drum ball mill level.

Prestack seismic data is imaged by calculating an individual reflectivity for each frequency in the seismic data. Then, a mean reflectivity is calculating over the individual reflectivities. A variance is calculated for the set of reflectivities versus frequency. A second variance is calculated for the upgoing wavefield, using the mean reflectivity. A spatially varying pre-whitening factor is then calculated, using the variance for the reflectivities and the variance for the upgoing wavefield. A reflectivity is calculated at each location, using the spatially varying pre-whitening factor.

The present invention relates to a method for rotation speed extraction through peeling off fault corresponding impact by employing a rolling ball track, belonging to the field of the fault diagnosis technology and the signal processing analysis technology. A charge amplifier is employed to perform amplification process of the original vibration signals, collected by an accelerometer, of a fault rolling bearing; the original vibration signals after the amplification process are subjected to the AR pre-whitening processing, an MED filtering method with the combination of a spectrum kurtosis is employed to enhance the fault impact feature in the pre-whitening signals, and the obvious vibration signals of the fault impact feature are obtained; the obvious vibration signals of the fault impact feature are subjected to Hillbert transformation, and enveloping signals are constructed; the impact feature position information in the enveloping signals of larger than a threshold is extracted; the rotation speed of a maintenance support is calculated; and the rotation speed of a rotating shaft is calculated according to the rotation speed of the maintenance support. The method provided by the invention is much simpler and better effective than other rotation speed extraction methods, can avoid the inhibition of the work condition for the installation of a collection device, and can save the rotation speed extraction cost.

Prestack seismic data is imaged by calculating an individual reflectivity for each frequency in the seismic data. Then, a mean reflectivity is calculating over the individual reflectivities. A variance is calculated for the set of reflectivities versus frequency. A second variance is calculated for the upgoing wavefield, using the mean reflectivity. A spatially varying pre-whitening factor is then calculated, using the variance for the reflectivities and the variance for the upgoing wavefield. A reflectivity is calculated at each location, using the spatially varying pre-whitening factor.

Duplex audio communications over a network use compressed audio data, with linear prediction coefficients (LPCs) and variances by which sample values differ from predictions. A adaptive echo canceller for a transceiver develops finite impulse response filter (FIR) coefficients characterizing an echo path between its local audio output and audio input. The received / decompressed audio data is applied to the FIR coefficients, and the predicted echo is subtracted from the uplink signal. Echo is detected as cross-correlation of the receive signal versus the uplink / send signal over time. In one embodiment, the cross-correlation is determined using a pre-whitened receive signal, obtained by adopting the variance values received over the network by the downlink Codec. Apart from the uplink Codec, no speech analysis filter or process is needed. The technique is apt for GSM, AMR and similar compressed audio communications.

A MIMO OFDM system includes a plurality of space-time encoders for encoding respective data blocks with independent space-time codes. The transformed data block signals are transmitted by a plurality of transmit antennas and received by a plurality of receive antennas. The received data is pre-whitened prior to maximum likelihood detection. In one embodiment, successive interference cancellation can be sued to improve system performance. Channel parameter estimation can be enhanced by weighting the channel impulse response estimates based upon a deviation from average.

The invention discloses a kurtosis-based variable-step-size self-adaptive blind source separation method, and aims to realize self-adaptation by judging distance between a solution of an algorithm and an optimal solution through kurtosis, adjusting step size on line, and continuously optimizing a separation matrix. The method specifically comprises the following steps of: 1, pre-whitening an observation signal; 2, iterating the separation matrix W by using the whitened signal; and 3, acquiring an optimal matrix to realize source signal separation. The kurtosis-based variable-step-size self-adaptive blind source separation method has the advantages that the step size is controlled according to the change of the kurtosis; the aim of the self-adaptation is fulfilled by judging the distance between the solution of the algorithm and the optimal solution through the kurtosis, adjusting the step size on line, and continuously optimizing the separation matrix; and a contradiction between convergence speed and steady-state errors in a blind source separation process is solved.

The application relates to a method for determining at least one updated filter coefficient of an adaptive filter (22) adapted by an LMS algorithm. According to the method, filter coefficients of a first whitening filter (25′) are determined, in particular filter coefficients of an LPC whitening filter. The first whitening filter (25′) generates a filtered signal. A normalization value is determined based on one or more computed values obtained in the course of determining the filter coefficients of the first whitening filter (25′). The normalization value is associated with the energy of the filtered signal. At least one updated filter coefficient of the adaptive filter (22) is determined in dependency on the filtered signal and the normalization value. Preferably, updated filter coefficients for all filter coefficients of the adaptive filter (22) are determined.

Duplex audio communications over a network use compressed audio data, with linear prediction coefficients (LPCs) and variances by which sample values differ from predictions. A adaptive echo canceller for a transceiver develops finite impulse response filter (FIR) coefficients characterizing an echo path between its local audio output and audio input. The received / decompressed audio data is applied to the FIR coefficients, and the predicted echo is subtracted from the uplink signal. Echo is detected as cross-correlation of the receive signal versus the uplink / send signal over time. In one embodiment, the cross-correlation is determined using a pre-whitened receive signal, obtained by adopting the variance values received over the network by the downlink Codec. Apart from the uplink Codec, no speech analysis filter or process is needed. The technique is apt for GSM, AMR and similar compressed audio communications.

The invention discloses an arbitrary array coherent source direction finding method under impact noise. The method comprises the following steps: establishing a sampling signal model; constructing a real array dynamic random weighting low-order covariance matrix; defining an interpolation transformation matrix T, and constructing a virtual array covariance matrix; obtaining a data covariance matrix and a noise covariance matrix after forward and backward space smoothing correction, and performing pre-whitening processing to obtain a dynamic random weighted low-order covariance matrix; estimating the number of information sources, performing feature decomposition on the dynamic random weighting covariance matrix, and determining a signal subspace and a noise subspace; constructing a spectrum estimation formula of a dynamic random weighted low-order covariance-spatial smoothing-MUSIC direction finding method, carrying out spectrum peak searching, finding the angle corresponding to the maximum value point, and outputting any array coherent source direction finding result. According to the method, the signal source incoming wave direction of any array can be effectively estimated, direction finding can be carried out under Gaussian noise, weak impact noise and strong impact noise, the decoherence performance is excellent, and the application range is wide.

The invention relates to an object recognition method based on a convolutional restricted Boltzmann machine combining Centering Trick. The method includes: structuring a training database, reconstructing an image, performing pre-whitening, using a CRBM model combining Centering Trick to perform feature extraction for a training set, transforming a 3-dimensional eigenmatrix into a one-dimensional eigenvector, using a Softmax classifier to classify the features, obtaining a test set, recombining and whitening the test set, computing the assumed value Si corresponding to output of a pool layer, and determining whether classification is correct or not according to a fact that whether Si is equal to the label of the test set or not. During a feature extraction process, a memory needed for computation is reduced and the computing speed is increased by making full use of two dimensionality of an image itself and the relations between pixels and also by using a CRBM model to achieve weight sharing. By using Centering Trick, the noise of inputting ofeach layer is reduced, and the accuracy of calculation and the stability of the model are improved.

The invention provides an evaluation method for marine controllable source electromagnetic exploration noise reduction effects, and concretely relates to an evaluation method for noise reduction effects of earth natural electromagnetic signal noise in marine controllable source electromagnetic exploration. A mean square relative error of a resistivity inversion result and a resistivity input valueare used as quantitative evaluation indexes of the noise reduction effects by combining MCSEM signal one-dimensional forward modeling, MT noise one-dimensional forward modeling, noise reduction processing and one-dimensional OCCAM inversion. MT noise can be generated by using a forward modeling means, and actual measured data of an MCSEM transmitter in a power-off time segment can be directly input to be the MT noise, and so the evaluation method can be ensured to have strong flexibility. Quantitative evaluation is carried out on the noise reduction effects of a method combining a short timewindow and pre-whitening filtering and the validity of the noise reduction effects is verified by using the evaluation method of the MT noise reduction effects provided by the invention. The noise reduction effect evaluation method provided by the invention has a wide application range, and fills the blank in the field of quantitative evaluation of noise reduction effects.

The invention discloses an adaptive echo cancellation method of a time-frequency memory sub-band proportion, and belongs to the technical field of adaptive electroacoustic echo cancellation. In view of the echo cancellation problem, the invention provides the adaptive echo cancellation method of thetime-frequency memory sub-band proportion. On one hand, in view of the instable properties of input echo signals, a sub-band structure is adopted to perform decorrelation on the input echo signals on the frequency domain, and an AP algorithm is adopted to perform pre-whitening from the time domain so as to acquire better echo cancellation performance; and on the other hand, a scale factor carrying time-frequency memory is used for resisting against the sparsity of an echo channel to acquire a higher convergence rate.

The invention discloses an optimized convolution automatic encoding network-based auroral image sorting method, and mainly aims at solving the problem that the existing technology is relatively low in auroral image sorting accuracy. The method comprises the following realization steps: 1, solving a saliency map of auroral images and extracting training samples on the basis of the saliency map; 2, carrying out pre-whitening on the training samples; 3, training an automatic encoding network AE; 4, solving the convolution self-encoding characteristics of the auroral images by utilizing the trained automatic encoding network; 5, carrying out mean pooling on the convolution self-encoding characteristics of the auroral images; and 6, inputting the pooled convolution self-encoding characteristics into a softmax sorter so as to sort the auroral images.

The invention provides a rolling bearing fault positioning method based on pre-whitening analysis. After a real cepstrum of a rolling bearing vibration signal is obtained, the non-zero quefrency containing nonstationary shock caused by white noise and bearing damage is reserved, the cepstrum component, corresponding to the quefrency expect for the non-zero quefrency, in the real cepstrum is set tobe zero, complex interference of the outside world is greatly lowered in the process, the signal-to-noise ratio of the rolling bearing vibration signal is increased advantageously, the envelope spectrum with the higher precision is obtained, thus the rolling bearing fault is effectively positioned, and the problem that in the prior art, when transmission systems of an aircraft and rail transit equipment is abnormal in operation, the fault type of a bearing cannot be positioned in an isolated mode can be solved.

Prestack seismic data is imaged by calculating an individual reflectivity for each frequency in the seismic data. Then, a mean reflectivity is calculating over the individual reflectivities. A variance is calculated for the set of reflectivities versus frequency. A second variance is calculated for the upgoing wavefield, using the mean reflectivity. A spatially varying pre-whitening factor is then calculated, using the variance for the reflectivities and the variance for the upgoing wavefield. A reflectivity is calculated at each location, using the spatially varying pre-whitening factor.

The invention belongs to the field of civil engineering structure health monitoring, and provides a dynamic non-Gaussian structure monitoring data exception identification method. The method comprises the steps of firstly, defining past and current observation vectors for monitoring data, and performing pre-whitening on the past and current observation vectors; secondly, building a statistical correlation model for the whitened past and current observation vectors, and obtaining dynamic whitened data; thirdly, dividing the dynamic whitened data into a system-related part and a system-unrelated part, and performing independent component analysis modeling on the system-related part and the system-unrelated part; and finally, defining two statistical quantities, determining control limits of the two statistical quantities, and when the statistical quantities exceed the control limits, judging whether an exception exists in the monitoring data. The non-Gaussianity and the dynamic characteristic of the structure monitoring data are considered at the same time, so that the exception in the data can be effectively identified based on the defined statistical quantities.

The embodiment of the invention, which relates to the wireless communication field, discloses a signal detection method and apparatus, thereby improving the detection performance of the signal detection algorithm. According to the embodiment of the invention, a receiver receives a wireless communication signal transmitted by a transmitter; the receiver carries out pre-whitening processing on the received wireless communication signal according to a non-ideal factor pre-whitening matrix, wherein the non-ideal factor pre-whitening matrix is obtained based on determination by the receiver at least according to a power of a non-ideal factor error vector; and then the receiver carries out signal detection on the wireless communication signal after pre-whitening processing. Therefore, the detection performance of the signal detection algorithm can be improved.

The application relates to a method for determining at least one updated filter coefficient of an adaptive filter (22) adapted by an LMS algorithm. According to the method, filter coefficients of a first whitening filter (25′) are determined, in particular filter coefficients of an LPC whitening filter. The first whitening filter (25′) generates a filtered signal. A normalization value is determined based on one or more computed values obtained in the course of determining the filter coefficients of the first whitening filter (25′). The normalization value is associated with the energy of the filtered signal. At least one updated filter coefficient of the adaptive filter (22) is determined in dependency on the filtered signal and the normalization value. Preferably, updated filter coefficients for all filter coefficients of the adaptive filter (22) are determined.

The invention consists of a filter, arithmetic unit and a self-adapting calculus device. The filter has a transform matrix device used in pre-whitening signal, and is used in restoring signal into a reduction signal. The arithmetic unit is used in comparing the reduction signal with an expected signal to get an error signal. The self-adapting calculus device is used in using error signal to adjust a filter coefficient of filter, and adopts normalized least mean square, NLMS algorithm.

The invention discloses a nonlinear digital self-interference cancellation device and method applied to a full-duplex communication transceiver. The nonlinear digital self-interference cancellation device applied to the full-duplex communication transceiver comprises a pre-whitening module, adaptive filters A and B and an adder; the pre-whitening module is used for pre-whitening a baseband waveform signal x(n) of a transmitter end and a linear mirroring component, a nonlinear component and a mirroring nonlinear component thereof; the adaptive filters A and B are used for performing iterative operations using a pre-whitening enhanced two-channel nonlinear complex number least mean square error adaptive algorithm according to the real parts or imaginary parts of an output signal of pre-whitening module and an output signal at the last iteration, obtaining an estimation of each main interference component in an observation signal d(n) of a receiver end, and eliminating each main interference component from d(n) and outputting; and the adder is used for taking the output of the adaptive filter A as the real part and the output of the adaptive filter B as the imaginary part, and addingthe real part and the imaginary part as a digital self-interference removed signal e(n) to output. The invention has a lower complexity.

The invention discloses a four-order tensor joint diagonalization algorithm for joint blind source separation of four data sets. The method comprises the following steps: S1, observation signal: pre-whitening the observation signals of the four data sets respectively; S2, target tensor: after the pre-whitening, constructing a set of mutual four-order cumulant tensors; S3, initializing a factor matrix; and S4, cost function convergence calculation: if the algorithm converges after once scanning, the calculation ends, if the algorithm still does not converge, then, the factor matrix obtained bythe present update is used as an initial value, and the next scanning is performed, a Jacobian rotation matrix is traversed and updated to update the factor matrix until convergence. According to thefour-order tensor joint diagonalization algorithm for joint blind source separation of four data sets, the algorithm is based on orthogonal rotation transformation, thereby an optimal solution in thesense of least squares can be obtained, and the algorithm is a J-BSS method for no more than four dataset signals.

A transform domain adaptive filter (TDAF) is provided. The transform domain adaptive filter includes a filter device, a computing device and an adaptive algorithmic device. The filter device has a transform matrix device for pre-whitening an input signal to obtain a pre-whitened signal. The filter device restores the pre-whitened signal to a restored signal. The computing device compares the restored signal with a reference signal to obtain an error signal. The adaptive algorithmic device adjusts a filter coefficient of the filter device. The adaptive algorithmic device adopts the normalized least mean square (NLMS) algorithm.