13 results about "Spectral density estimation" patented technology

The invention provides a voice enhancement and high-precision recognition method and system in a complex environment, and relates to the technical field of voice processing, and the method comprises the steps: collecting a time domain signal in an off-road parking sentry box environment for preprocessing, detecting a mute segment signal in a standard time domain signal for noise power spectral density estimation, and obtaining a noise power spectral density value; a reverberation parameter is obtained by combining voice onset information and noise spatial correlation estimation, prediction is performed by using a deep neural network model, voice masking is applied to microphone array signals to perform enhancement processing, adaptive feature extraction is performed on time domain enhanced voice signals, and a voice signal is obtained. And performing high-precision recognition on the voice adaptive feature sequence based on an acoustic model and a language model, and outputting a target recognition text. The technical problems of poor voice signal quality and low recognition accuracy in a complex noise environment in the prior art are solved. The technical effects of improving the voice signal quality and the recognition accuracy and realizing clear, accurate and real-time voice interaction are achieved.

The invention discloses a large-span roof structure wind-induced response fast algorithm considering high-order modal quasi-static response, and the algorithm comprises the following steps: S11, building a finite element model of a large-span roof structure, and determining the modal damping ratio of the large-span roof structure according to the design condition, so as to obtain the structural dynamic characteristics of the large-span roof structure; step S12, determining a random wind load acting on the large-span roof structure through a wind tunnel test, and calculating a wind load Fourier transform function of a truncation function corresponding to the random wind load; step S13, according to the dynamic characteristics of the structure and the wind load Fourier transform function, performing calculation by adopting a modal acceleration method, obtaining a Fourier transform function of the structure displacement response; and S14, according to the Fourier transform function of the structure displacement response, performing power spectral density estimation by adopting a periodogram method to obtain a power spectral density matrix of the structure displacement response, and performing integration on the power spectral density matrix of the structure displacement response to obtain a displacement response root-mean-square of the large-span roof structure. According to the method, quasi-static response of a high-order mode is considered, better convergence is achieved compared with a traditional method, and under the condition of the same precision, calculation time and memory occupation are relatively small.

The invention discloses a pulling and lifting cable frequency detection and cable force measurement method and system based on a millimeter wave radar, and belongs to the field of pulling and lifting cable force measurement. The method comprises the steps that vibration signals are collected through the millimeter wave radar to obtain complex data streams, time-frequency analysis is conducted on the complex data streams, and an interference model is established; the method comprises the following steps: reconstructing a noise-reduced vibration signal by adopting adaptive filtering, extracting candidate vibration frequencies through power spectral density estimation and spectral peak detection, calculating corresponding stability scores, and obtaining a first verification factor used for evaluating wind disturbance and a second verification factor used for evaluating vibration frequency deviation; according to the method, through multi-source sensing data fusion and self-adaptive signal processing, complex environment interference is effectively overcome, non-contact high-precision cable force measurement is achieved, and the method has the outstanding advantages of being high in anti-interference capacity, high in measurement reliability and the like.

A fiber optic sensing cable located along a side of a paved road and runs parallel to a driving direction is monitored by distributed fiber optic sensing (DFOS) using Rayleigh backscattering generated along the length of the optical sensor fiber cable under dynamic vehicle loads. The interaction of vehicles with roadway locations exhibiting distressed pavement generates unique patterns of localized signals that are identified / distinguished from signals resulting from vehicles operating on roadway exhibiting a smooth pavement surface. Machine learning methods are employed to estimate an overall road surface quality as well as localizing pavement damage. Power spectral density estimation, principal component analysis, support vector machine (SVM) combined with principal component analysis (PCA), local binary pattern (LBP), and convolutional neural network (CNN) are applied to develop the machine learning models.

According to the logarithmic scale adaptive multi-window power spectral density estimation method provided by the invention, the conicity degree is adjusted through the equilibrium relationship between the deviation and the variance, the problems of low resolution and high algorithm complexity of the existing adaptive sine window algorithm are solved, and the improvement of the adaptive sine window algorithm is realized. The method comprises the following steps: firstly, initializing conicity degrees, setting the conicity degrees at all frequencies as fixed values, and calculating a power spectral density initial value; secondly, cone degree gradient processing is carried out on low-frequency signals, the cone degree is smoothed through filters with different intensities at different frequency bands, and then the optimal cone degree of the whole frequency domain is obtained; and finally, carrying out iterative operation on the conicity degree and the power spectrum density to obtain an optimal power spectrum density estimation value. Compared with an LPSD algorithm, an adaptive sine window algorithm and an LISA-LPSD algorithm, the method provided by the invention has the advantages of high spectrum resolution, low algorithm complexity and the like.

The present invention discloses a signal recognition method based on broadband spectrum detection. The FPGA is used to perform power spectrum estimation and signal detection on broadband IQ data. After obtaining the signal frequency point parameters, the DDC module is configured to complete digital downconversion. The power spectral density estimation and noise floor estimation of narrowband IQ signals are performed on the upper computer. Preliminary classification is carried out by calculating the ratio parameter C1 of the maximum value to the second maximum value of the spectrum. When C1 is greater than the threshold, the AM and single-carrier signals are discriminated by combining the spectrum feature parameter C2 and the amplitude spectrum feature parameter C3. When C1 is less than or equal to the threshold, a double-threshold algorithm is used for signal detection and the merging and screening in the multi-signal scenario are completed. Classification is carried out based on the normalized zero-centered instantaneous amplitude spectrum density maximum value parameter C4. When C4 is less than the threshold, the FM, MSK, and FSK signals are discriminated. When C4 is greater than or equal to the threshold, the BPSK, QPSK, 8PSK, and 16QAM signals are discriminated. The present invention adopts a multi-parameter joint decision strategy, which significantly improves the accuracy and real-time performance of signal recognition.

The invention discloses a multi-domain fusion super-multilevel attention evaluation method and system based on electroencephalogram data, and the method comprises the following steps: collecting an attention electroencephalogram signal of a test person through an electroencephalogram signal collection device, carrying out the filtering processing, and obtaining a processed electroencephalogram signal; carrying out attention state division based on the processed electroencephalogram signal to obtain a state division result; performing time-frequency decomposition on the processed electroencephalogram signal by using fast Fourier transform to obtain a weighted time-domain signal; performing power spectral density estimation on the processed electroencephalogram signals by using a Welch method to obtain a multi-channel fusion image; constructing a parallel model structure, and training the parallel model structure based on the weighted time domain signal, the multi-channel fusion image and the state division result to obtain an attention evaluation model; and collecting attention electroencephalogram signals of the to-be-evaluated person, and evaluating the attention state of the to-be-evaluated person based on the attention evaluation model to obtain an attention evaluation result.

The invention relates to the technical field of signal processing, and provides an airborne early warning radar main lobe unintentional interference resisting method and device based on frequency domain filtering. The method comprises the following steps: performing interference spectrum estimation through adaptive beam forming and a frequency spectrum estimation algorithm based on a maximum likelihood criterion to obtain a power spectrum density estimator of each frequency point; determining a frequency range of the interference signal according to the power spectral density estimator of each frequency point; a plurality of channels are constructed according to the frequency range of the interference signal, and each channel corresponds to an interference filter; in each channel, filtering the interference signal by using an interference filter to obtain a non-interference signal of each channel, and for color noise caused by filtering, performing noise compensation on the non-interference signal of each channel to obtain a channel output signal of each channel; and selecting the channel output signal with the minimum power as a final output signal. According to the method, main lobe unintentional interference suppression is completed, and the target detection performance of the airborne radar is improved.

The present invention relates to the technical field of signal processing, and provides a method and device for an airborne early warning radar to resist main lobe unintentional interference based on frequency domain filtering. The method includes: performing interference spectrum estimation through an adaptive beamforming and frequency spectrum estimation algorithm based on the maximum likelihood criterion to obtain the power spectral density estimation quantity of each frequency point; determining the frequency range of the interference signal according to the power spectral density estimation quantity of each frequency point; constructing a plurality of channels according to the frequency range of the interference signal, and each channel corresponds to an interference filter; in each channel, using the interference filter to filter out the interference signal to obtain the interference-free signal of each channel, and performing noise compensation on the interference-free signal of each channel for the color noise caused by filtering to obtain the channel output signal of each channel; selecting the channel output signal with the minimum power as the final output signal. The present invention completes the suppression of main lobe unintentional interference and improves the target detection performance of the airborne radar.

The invention provides a nut flavor intelligent perception evaluation method, device and equipment and a storage medium, frequency domain feature extraction is carried out on electroencephalogram data through a Welch method, power spectral density estimation is carried out according to the extracted frequency domain feature, and the frequency domain feature power spectral density is obtained; for a plurality of preset brain wave segments of different brain regions, calculating a power spectral density average value of the plurality of preset brain wave segments in each brain region according to the frequency domain characteristic power spectral density; and constructing a nut flavor sensing electroencephalogram feature contour based on the average value of the power spectrum densities of the different brain regions. Compared with the prior art that a modern instrument can only measure the content of volatile components, the method constructs a perceived electroencephalogram characteristic contour according to the average value of the power spectrum density of different brain regions, realizes intelligent perception evaluation based on electroencephalogram physiological signals, and provides an objective and quantitative analysis tool for food sensory science.

This invention relates to brain-computer interfaces, specifically to a brain-computer interface-based emotion and stress management system. The control unit acquires the user's multimodal signals using a non-invasive multimodal signal acquisition device and preprocesses the EEG signals using an EEG signal processing module. The control unit then estimates the power spectral density of the preprocessed EEG signals using a power spectral density estimation module to obtain the power spectral density of the EEG signals. A relaxation level acquisition module obtains the user's relaxation level based on the power spectral density of the EEG signals, a coefficient of variation acquisition module obtains the coefficient of variation for the corresponding frequency band based on the power spectral density of the EEG signals, and a Schumann resonance energy index acquisition module obtains the Schumann resonance energy index corresponding to the first and second harmonics based on the power spectral density of the EEG signals. The technical solution provided by this invention overcomes the shortcomings of existing technologies, such as the difficulty in accurately identifying the user's emotional state and the inability to provide personalized real-time intervention.

The present invention discloses a spectrum-sensing-based adaptive channelization detection method, belonging to the field of electronic reconnaissance, which performs spectrum sensing, spectrum analysis, channel matching, detection, and parameter estimation on electromagnetic signals and outputs detection results. First, the electromagnetic signals in the monitored frequency band are successively subjected to AD sampling and short-time Fourier transform (STFT) to obtain spectrum-sensing data; the spectrum-sensing data is input for real-time energy spectral density estimation to generate a mean spectrum and a maximum spectrum; the frequency bands where the signals are distributed are subjected to spectrum analysis through the mean spectrum and the maximum spectrum to generate channel adaptation parameters; the digital channelization structure is adaptively adjusted according to the channel adaptation parameters to achieve the optimal matching between the signal bandwidth and the channelization bandwidth, and the input electromagnetic signals are subjected to adaptive channelization detection and parameter estimation to complete the output of the final detection results.

The present invention discloses a fast algorithm for wind-induced response of long-span roof structures considering the quasi-static response of high-order modes, comprising the following steps: Step S11, establishing a finite element model of the long-span roof structure and determining the modal damping ratio of the long-span roof structure according to the design conditions to obtain the structural dynamic characteristics of the long-span roof structure; Step S12, determining the random wind load acting on the long-span roof structure through a wind tunnel test and calculating the Fourier transform function of the wind load corresponding to the truncation function of the random wind load; Step S13, calculating by using the modal acceleration method according to the structural dynamic characteristics and the Fourier transform function of the wind load to obtain the Fourier transform function of the structural displacement response; Step S14, performing power spectral density estimation by using the periodogram method according to the Fourier transform function of the structural displacement response to obtain the power spectral density matrix of the structural displacement response, and integrating the power spectral density matrix of the structural displacement response to obtain the root mean square of the displacement response of the long-span roof structure. In the present invention, the quasi-static response of high-order modes is taken into account, which has better convergence than the traditional method, and under the condition of the same accuracy, the calculation time and memory occupation of the present invention are relatively small.