33 results about "Spectral noise" patented technology

The invention provides a large-model cross-modal collaborative understanding method and device. The fusion efficiency and the understanding capability of multi-modal information can be improved. The large-model cross-modal collaborative understanding method comprises the following steps: preprocessing visual data, language data and sound data to obtain preprocessed visual data, language data and sound data; wherein the preprocessing comprises the steps of performing adaptive size adjustment and normalization on visual data, performing word segmentation and dynamic truncation on language data, and performing band-pass filtering and spectral noise reduction on sound data; extracting visual features, language features and sound features based on the preprocessed visual data, language data and sound data; on the basis of an adaptive mapping network and a mixed granularity cross-modal attention mechanism, performing feature alignment on the visual features, the language features and the sound features to obtain aligned feature vectors; and based on a dynamic routing architecture, fusing the aligned feature vectors to generate a unified multi-modal representation.

The invention relates to the technical field of artificial intelligence and data processing, and discloses an energy storage equipment operation state monitoring method based on Internet of Things acquisition, which comprises the following steps: constructing a vibration data sample; performing state labeling on the vibration data sample to form a labeling database; performing sliding mean filtering and linear interpolation processing on the vibration data sample; carrying out adaptive spectrum noise suppression by adopting a dual-threshold wavelet packet noise reduction function; carrying out frequency band division by adopting a weighted multi-scale spectral entropy feature enhancement method; constructing a deep neural network model for classifying the running state of the energy storage equipment, and training the deep neural network model by using a labeling database; and inputting preprocessed vibration data acquired in real time into the trained deep neural network model, and performing state identification and fault alarm according to a category probability prediction vector output by the model. The running state of the energy storage equipment can be accurately monitored in real time, potential faults can be found in time, and measures can be taken.

Systems and methods are provided for hysteretic mode boost voltage control for reduced broad spectrum noise emission.

The invention discloses a Raman spectrum noise reduction method based on an MSC-U-Net algorithm, and the method comprises the steps: S1, measuring Raman spectrums of a plurality of substances with different concentrations under the same condition, and respectively measuring a plurality of spectrums for each concentration of each substance, so as to obtain an original spectrum data set containing noise; s2, performing normalization preprocessing on all the collected spectrums, dividing a data set into a training set, a verification set and a test set, and constructing a data set required by a noise reduction model; s3, averaging the collected original spectrums according to a plurality of spectrums of the same substance and the same concentration to obtain ideal spectrums without noise; and S4, constructing a Transform-Attention U-Net model, inputting the training set into the model to carry out parameter learning, and carrying out iterative optimization according to a loss function to obtain a spectral noise reduction model. According to the method, noise signals in the Raman spectrum can be effectively removed, meanwhile, effective signals are not lost, and the signal-to-noise ratio of the Raman spectrum is increased.

The application discloses an image processing method, a storage medium and a computer terminal. The method comprises the following steps: obtaining an initial spectral image, wherein the initial spectral image contains a target object, and the initial spectral image is composed of image information and first spectral information of a plurality of spectral channels; superimposing spectral noise on the initial spectral image to obtain a target spectral image; and identifying the target spectral image to determine the category of the target object. The application solves the technical problem of low accuracy of spectral image identification caused by the influence of other spectral signals in the related art.

The invention discloses a mosaic multispectral image quality optimization method and system based on spectrum noise reduction, and the method comprises the steps: obtaining an original snapshot mosaic multispectral image, and carrying out the radiation correction and spectrum correction of the original snapshot mosaic multispectral image, and obtaining a corrected snapshot mosaic multispectral image; based on fast Fourier transform, converting the corrected snapshot type mosaic multispectral image into a spectrogram; detecting and positioning a noisy point region on the spectrogram to obtain the noisy point region; removing noise corresponding to the noisy point area from the spectrogram to obtain a spectrogram with the noise removed; and on the basis of inverse Fourier transform, the spectrogram with the noise removed is transformed into an optimized mosaic multispectral image. For the snapshot type mosaic multispectral image, grid noise removal can be rapidly achieved, meanwhile, the influence of exposure conditions is overcome, and quality optimization of the snapshot type mosaic multispectral image is achieved on the premise that a large-scale data set training model does not need to be manufactured.

The invention relates to a pollutant detection method and system based on machine learning, and relates to the technical field of pollutant detection.The method comprises the steps that original spectral data are collected and denoised through a Savitz-golay smoothing filter, parameters are dynamically adjusted, and stable spectral data with pollutant core characteristics reserved are obtained; the method can effectively solve the technical problem that low-concentration interval heterovariance noise masks weak features of left censored data, and avoids the defect of excessive smoothness or insufficient denoising of a traditional denoising method. Secondly, in combination with a blank spectrum noise optimization detection limit and a quantification limit, effective features are extracted by utilizing machine learning, and incomplete observation characteristics of left-censored data can be adapted, so that the problem of systematic bias caused by data distribution hypothesis distortion of an existing algorithm is solved, and the defect that a low-concentration risk early warning window is neglected in a traditional method is overcome; finally, filter parameters are analyzed and optimized through a loss function, a detection result is finally output, and detection limit heterogeneity caused by instrument drifting and working condition fluctuation can be dealt with.

This invention relates to the fields of artificial intelligence and data processing technology, and discloses a method for monitoring the operating status of energy storage devices based on IoT data acquisition. The method includes: constructing vibration data samples; labeling the vibration data samples to form a labeled database; performing moving average filtering and linear interpolation on the vibration data samples; using a dual-threshold wavelet packet denoising function for adaptive spectral noise suppression; using a weighted multi-scale spectral entropy feature enhancement method for frequency band division; constructing a deep neural network model for classifying the operating status of energy storage devices, and training the deep neural network model using the labeled database; inputting the preprocessed vibration data acquired in real time into the trained deep neural network model, and performing status identification and fault alarm based on the class probability prediction vector output by the model. This invention can monitor the operating status of energy storage devices in real time and accurately, promptly detect potential faults, and take appropriate measures.

The invention discloses a high-stability infrared lighting system and method based on an integrated reflection halogen lamp. Comprising a modularized light source, a spectrum purification unit, a thermal management unit, a monitoring and control unit, a light output unit and an optional light path expansion unit. The modularized light source adopts a light source engine module design capable of being quickly plugged and unplugged, and is integrated with a reflecting halogen lamp and a modularized lamp holder; the spectrum purification unit realizes spectrum purification through an infrared band-pass optical filter, and the heat management unit performs cooperative heat dissipation on the reflection halogen lamp and the infrared band-pass optical filter through a heat dissipation fan; and the monitoring and control unit realizes brightness closed-loop control, temperature closed-loop control and light source service life management and early warning through feedback of a light intensity sensor and a temperature sensor. The problems that in the prior art, spectral noise is large, output stability is poor, thermal management is difficult and maintenance is inconvenient are solved from the source, an infrared illumination solution which is high in strength, high in uniformity, stable for a long time and easy to maintain is provided, and the system is suitable for semiconductor wafer infrared transmission detection and other precise optical detection scenes.

The application describes a method and apparatus for decoding voice speech. The method may include receiving encoded data including linear predication coding (LPC) coefficients, an indication of a fixed codebook (FCB), and an indication of an adaptive codebook (ACB). The encoded data may be decoded into a excitation signal. A spectral analysis may performed on the excitation signal. A time envelope of the excitation signal may be applied to a white noise signal and filtered to obtain a complementary noise signal. The complementary noise signal may be summated with the excitation signal to recover a modulated signal. The application also describes a method and apparatus for decoding unvoiced speech.

The application provides a bridge cable force automatic tracking method based on frequency domain energy optimization recombination. During bridge operation monitoring, aiming at the difficulty in automatic picking of peaks caused by the noise pseudo-peak of cable acceleration signal spectrum and the time-varying distribution difference of modal energy in the frequency domain, firstly, the normalized fast Fourier transform algorithm is used to convert the cable vibration time history into the frequency domain in sections; then the obtained multiple frequency spectra are subjected to mean smoothing pretreatment to obtain the smoothed spectrum with the pseudo-peak noise removed. Then, according to the logarithmic processing of the smoothed spectrum, the optimization recombination of the modal energy of each order in the frequency domain is realized to improve the robustness of the automatic picking of peaks; finally, according to the setting of the reasonable interval of the automatic picking of peaks in the frequency domain based on the theoretical fundamental frequency of the cable, and the limitation of the normalized frequency domain energy, the frequency domain sliding window peak searching algorithm and the vibration string theory formula are combined to realize the automatic stable tracking of the bridge cable force. The application can solve the technical problem that the cable force calculation of the traditional method is not accurate enough.

The invention provides an intelligent electric control inclined dome near-infrared light source calibration system. The system comprises a rack, a console, a vertical electric lifter and a dome light source. The inclined dome light source is an integrally-formed cone, and four halogen lamp cups are evenly distributed on the inclined face and used for guaranteeing the uniformity of the light source. The inclined dome light source is installed on a horizontal moving sleeve frame, the horizontal position can be adjusted, a horizontal moving sleeve is connected with a vertical electric lifter, and the electric lifter is adjusted by measuring the height of a detection sample through a distance measuring sensor so as to adapt to the detection environment. A voltage-stabilized power supply is matched to prevent spectrum drift of the light source, and the controller adjusts the power of the light source and the distance between the light source and the sample according to signals of the distance measuring sensor. The method is used for solving the influence of a near-infrared detection system on detection of surface dazzle light, mirror reflection and unnecessary shadows of crops with different sizes, ensuring the uniformity of a light source and reducing spectral noise interference caused by uneven illumination.

The invention relates to a method for monitoring the integrity of a high-frequency (HF) path in a radar sensor, wherein the method comprises: measuring isolation between different radar channels to detect interference with an electromagnetic connection that indicates an assembly connection fault, and / or analyzing a noise level in the radar signal spectrum to detect an increase in spectral noise that indicates an assembly connection fault.

The invention discloses an LAI prior information-based vegetation biochemical parameter inversion method, which comprises the following steps of: sampling parameters according to an LAI value range, simulating a spectrum by using a vegetation radiation transfer model, and constructing data sets D10 and D20 containing LAI prior information; re-sampling the simulated spectrums in the D10 and the D20 to a set wave band, introducing spectral noise to obtain a lookup table D1 and a data set D21, adding uncertainty to LAI prior information in the D21 to obtain a data set D2, calculating the similarity of parameters in the D1 and the D2, screening candidate LAI parameter combinations from the D1, and forming a candidate solution set with the corresponding simulated spectrums; and selecting the optimal solution of the vegetation biochemical parameters in the D2 from the candidate solution set according to the root-mean-square errors of the spectrum in the candidate solution set and the spectrum to be measured on the plurality of LCC sensitive wavebands. The method can solve the problem that an existing leaf biochemical parameter inversion method is too highly dependent on LAI prior data accuracy.

The application provides a LiDAR point cloud classification method based on double-flow interaction and multi-scale mixed attention, relates to the technical field of hyperspectral laser radar point cloud processing, and comprises the following steps: spatial geometric features and spectral features are extracted in parallel through a double-flow interaction encoding module; a bidirectional guiding mechanism is used to realize cross-modal feature complementation; a multi-scale spectral-geometric mixed attention module is designed; spectral channel attention and multi-level K-neighbor-based geometric spatial pyramid attention are used to strengthen the global semantics and key geometric details of deep features; a semantic-spectral contrast loss function is introduced to improve the discrimination ability of the model for homogenous and heterogeneous objects by mining difficult example samples; and finally, the hyperspectral LiDAR point cloud is classified by using the trained model. The application can effectively suppress spectral noise interference, enhance the classification precision of the boundaries of ground objects, and is suitable for the classification of hyperspectral LiDAR point clouds in complex remote sensing scenes.

The application discloses a high-precision spectral noise elimination method and system based on TDLAS technology and relates to the technical field of gas concentration optical detection. The method obtains the center frequency and half-width parameter of a Lorentz model by fitting the standard gas sampling data, determines the adaptive cutoff frequency according to the center frequency and the half-width parameter, and performs digital low-pass filtering on the to-be-detected gas sampling data. The filtered data is divided into an absorption interval and two non-absorption intervals. The local phase information of the interference fringes in the non-absorption intervals is extracted and weighted third-order polynomial fitting is performed to obtain the baseline of the whole region. The application extracts the local phase of the interference fringes in the non-absorption intervals in real time and converts the local phase into the basis for weighted fitting. The processing link is constructed by combining adaptive filtering and constrained Lorentz fitting. The phase extrapolation compensation is used to eliminate the residual error of the interference fringes caused by temperature fluctuation, the spectral absorption area more accurately reflects the real absorption strength, and thus the precision of low-concentration gas detection is improved.

A Coptis chinensis moisture content analysis method based on heterogeneous sensor data fusion. The analysis method comprises the following steps: respectively collecting spectra of Coptis chinensis samples by using two types of spectrometers; gathering the collected spectra of the Coptis chinensis samples to form a sample set, and removing outliers to obtain a calibration set and a validation set; preprocessing the spectra, and selecting an optimal processing algorithm of the spectrometers to reduce spectral noise and simplify background spectral information; and performing low-level data fusion on spectra data, stitching spectral bands of near-infrared spectra and infrared spectra, so as to obtain moisture content prediction results, establishing a heterogeneous data fusion model, and on the basis of RMSE values and RPD values, evaluating the model and the prediction results, so as to obtain accurate analysis results.

The invention relates to the technical field of soil detection, in particular to a soil harmful substance residue detection method, device and system, and the method comprises the following steps: comparing the difference between the spectral intensity under any channel and the distribution condition of all fluorescence intensities in the channel section where the channel is located; determining a spectral noise confidence coefficient by combining the mutation probability of the fluorescence intensity under any channel in each channel interval so as to screen out a spectral peak; and determining an asymmetric factor based on the difference between each spectral peak in the baseline adjacent region of each channel interval and all fluorescence intensity distribution in the corresponding channel interval and the spectral noise confidence coefficient, and carrying out baseline correction and denoising on the spectrum detected each time by adopting a denoising algorithm for soil harmful substance residue detection. According to the method, the asymmetric proportion in the BEADS algorithm is dynamically optimized, spectral noise is effectively inhibited, trace spectral peaks are reserved, and the precision of soil harmful substance residue detection is improved.

The invention relates to a method for monitoring the integrity of a high-frequency (HF) path in a radar sensor, the method comprising: measuring an isolation between different radar channels in order to identify an influence on an electromagnetic connection which indicates a mounting connection fault, and / or analyzing a noise level in the radar signal spectrum in order to identify an increase in a spectral noise which indicates a mounting connection fault.

The invention provides a lightning whistle sound wave identification method and system based on multi-scale time-frequency fusion, and the method comprises the steps: carrying out the adaptive window length segmentation of a broadband very-low frequency sequence, calculating short-time Fourier transform under a plurality of time-frequency resolutions, carrying out the two-dimensional splicing of time and scale to construct a composite time-frequency image, giving consideration to long-time evolution and short-time details, and carrying out the recognition of lightning whistle sound waves. A self-supervised gating attention noise suppression mechanism is introduced, noise statistical characteristics are learned from historical observation, a spectrogram noise mask is generated, gating fusion is performed on a feature layer, self-adaptive suppression of non-stationary noise is realized, and whistle trace features are highlighted. And carrying out target detection on the enhanced composite spectrogram by adopting an improved lightweight YOLO network, and outputting a time-frequency boundary and main characteristic parameters of a whistle event. According to the method, efficient automatic identification and fine statistics of lightning whistle events can be realized in large-scale very-low-frequency observation data, and technical support is provided for related applications such as space environment monitoring based on lightning whistles.

The invention discloses a seismic data low-frequency compensation method, system and device and a medium, and the method comprises the steps: S1, constructing an expected wavelet frequency spectrum according to a frequency spectrum range of to-be-compensated seismic data and an effective low-frequency signal range, and enabling a constructed objective function to be s.tJ = 0; in the formula, f is an expected wavelet and is a modal coefficient, and ui is a modal component; s2, carrying out wavelet extraction on the seismic data to be compensated, and carrying out variational mode decomposition on the extracted wavelet; s3, compensating the modal coefficient of the low-order modal component according to the target function, and reconstructing the compensated modal coefficient and the decomposed modal coefficient into a new wavelet; and S4, performing convolution processing based on the reconstructed wavelets and the original data to obtain seismic data after low-frequency compensation. According to the method, the low-frequency band energy compensation degree can be quantified, after low-frequency compensation is carried out on the seismic data, the frequency spectrum noise is low, no sudden change condition exists, and the method is suitable for being applied and popularized in practice.

A computer-implemented procedure, when executed by data processing hardware, causes the data processing hardware to perform operations. These operations include receiving an audio signal from a sensor array at a unified post-filter, converting the audio signal into a short-time Fourier transform (STFT) domain using a conversion function, determining a speech presence probability based on the converted audio signal, and determining a noise smoothing factor based on the speech presence probability. The operations also include estimating a spectral noise power density based on the noise smoothing factor across the unified post-filter, estimating a control vector of a desired source across the unified post-filter, and generating a direction-based mask and a coherence-based mask across the unified post-filter.The operations also include generating a residual echo spectrum estimate based on the direction-based mask and the coherence-based mask, and setting one or more spectral shaping factors based on the residual echo spectrum estimate.

The application relates to a parameter determination method and device of a micro-perforated panel for spectrum noise reduction of intelligent electrical appliances, applied to the field of sound absorption and noise reduction, wherein the method comprises the following steps: determining a frequency attenuation curve of the micro-perforated panel based on initial parameters of the micro-perforated panel; obtaining original spectrum data of an intelligent electrical appliance to be reduced in noise; determining a filtering noise reduction amount according to the original spectrum data and the frequency attenuation curve; and predicting target parameters corresponding to a predicted noise reduction amount meeting a preset noise reduction threshold range in the micro-perforated panel based on a preset Bayesian optimization strategy and by using a target noise reduction prediction model. The target noise reduction prediction model is obtained by training a preset regression model based on the filtering noise reduction amount and the initial parameters of the micro-perforated panel, and is used for predicting filtering noise reduction amounts corresponding to different micro-perforated panel parameters. Through the application, the accuracy and efficiency of micro-perforated panel parameter design are improved.

This invention belongs to the interdisciplinary field of biomedical engineering and artificial intelligence-assisted diagnosis. It discloses a training method for a SERS spectral classification prediction model and its application. The disclosed SERS spectral classification prediction model training method combines CatBoost feature selection and deep learning. First, surface-enhanced Raman spectroscopy (SERS) data of serum from different categories of subjects is collected. After baseline removal, filtering, and normalization preprocessing, CatBoost gradient boosting algorithm is used to evaluate feature importance, considering the high-dimensional redundancy of the spectral data, and to select a subset of discrete feature bands containing key biomarker information. This feature subset is then input into a one-dimensional convolutional neural network model for training. The model constructed by this invention possesses deep feature mining capabilities, and the biological interpretability of the decisions is verified through SHAP analysis. It effectively solves the problems of large spectral noise interference and difficulty in extracting weak pathological features in traditional methods. For example, it can be used for the auxiliary diagnosis of coronary heart disease, achieving non-invasive, rapid, and high-precision classification and diagnosis of coronary heart disease and its subtypes.

Systems and methods are provided for hysteretic mode boost voltage control for reduced broad spectrum noise emission.

A test and / or measurement system and method are described. The test and / or measurement system includes a noise circuit, a first port, a second port, at least one receiver circuit, and at least one processing circuit. The noise circuit is configured to generate a noise signal having a spectrum extending over a predetermined frequency band of interest. The noise circuit is connected to a first port that is connectable to a first DUT port of a device under test such that a noise signal can be fed to the device under test. The second port is connectable to a second DUT port of the device under test in order to receive the test signal from the device under test, and the at least one receiver circuit is connected to the second port in order to receive the test signal. The at least one receiver circuit is configured to digitize the test signal to obtain a digitized test signal. The at least one processing circuit is configured to transform the digitized test signal to a frequency domain, thereby obtaining a transformed test signal having a plurality of frequency intervals.

The invention relates to the technical field of spectrum optimization, in particular to a near infrared spectrum characteristic wavelength optimization method for quantifying sensory quality of Pangao crisp melon, which comprises the following steps of: extracting average spectrum data of three parts of spectrum regions by taking near infrared spectrum regions at the head, the middle and the tail of a Pangao crisp melon sample as sensitive intervals; the method comprises the following steps of: preprocessing acquired near-infrared data by adopting Savitzky-Golay derivative + smoothing, and eliminating background information and spectral noise of an instrument; according to the method, a guided shrinkage two-stage optimization algorithm, namely a whale optimization-iterative wavelength combination optimization (WO-IWCO) method, is developed, sampling is performed by adopting a cyclic binary matrix sequence, a frequency spectrum interval is shortened, sensitive wavelengths are further compressed, and optimal characteristic wavelength variables related to the sensory quality of the Pangao crisp melons are more accurately screened out, so that the optimal characteristic wavelength variables related to the sensory quality of the Pangao crisp melons are more accurately screened out, and the optimal characteristic wavelength variables related to the sensory quality of the Pangao crisp melons are more accurately screened out, and the optimal characteristic wavelength variables related to the sensory quality of the Pangao crisp melons are more accurately screened out, and the optimal characteristic wavelength variables related to the sensory quality of the Pangao crisp melons are obtained.

The invention relates to the technical field of circuit testing, in particular to an aging test method for a timekeeping circuit of an electric energy meter based on multi-dimensional operation data. The specific implementation process comprises the following steps: acquiring electric energy meter time keeping circuit data, and reconstructing the data into a frequency domain texture energy spectrum diagram; spectrum noise particle size distribution characteristics are extracted, quartz crystal oscillation items are introduced to calibrate spectrum gravity center micro-drift, and circuit characteristic fingerprints are formed; obtaining aging texture ghosts through a synchronous harmonic stripping mechanism, and inputting the aging texture ghosts into a stress flux integration model to quantify a geometric erosion rate; and performing time sequence extrapolation to simulate an evolution trajectory according to the geometric erosion rate, and calculating and outputting a performance erosion index and a residual life prediction result. According to the invention, by using a quartz crystal oscillation item and a synchronous harmonic stripping mechanism, reversible temperature drift is effectively decoupled, environmental transient interference is separated, the reliability of evolution trajectory simulation under complex and real working conditions is ensured, and the accuracy of prospective early warning and residual life evaluation of the time keeping circuit of the electric energy meter is improved.

The embodiment of the invention provides a method, a circuit and a system for suppressing electromagnetic noise of a DC brush motor. The method comprises the following steps: configuring at least two filtering units with different frequency response characteristics; arranging at least two filtering units on a noise propagation path of the DC brush motor; noise components of different sub-frequency bands in the wide-spectrum noise generated by the direct-current brush motor are suppressed through at least two filtering units; wherein the synergistic effect of the at least two filtering units realizes suppression of electromagnetic noise in a preset full frequency band. The method is used for improving the denoising accuracy of the brush motor.

The application discloses a dynamic laser Doppler spectrum correction method, and the traditional laser Doppler spectrum correction method only suppresses fixed position noise of a spectrum, the position of Doppler spectrum noise corresponding to different laser velocimeter systems is not necessarily the same, therefore, the method is not universal, the suppression effect of the method is poor, and the spectrum noise is still unstable, thereby causing interference to spectrum resolution of the velocimeter system, and therefore, the application provides a laser Doppler spectrum correction method based on power spectrum analysis.The application performs signal suppression based on power spectrum analysis, has good universality, the spectrum noise suppression effect is obvious, the spectrum noise is stable, and the application lays a foundation for accurately extracting a Doppler signal in subsequent spectrum resolution.