57 results about "Reference noise" patented technology

The invention discloses a pattern recognition type metal pipeline anticorrosive coating quality detection device and system, belongs to the technical field of sensing monitoring, and aims to solve the problems of low detection precision and difficulty in positioning of damaged points of a metal pipeline anticorrosive coating under the interference of a complex environment. The method comprises the following steps: generating an environmental interference spectrum feature library through full-range frequency band pre-scanning, dynamically establishing a reference noise level, screening candidate frequencies of detection signals, and switching the frequencies as required to guarantee signal quality; synchronously acquiring a potential gradient and an electromagnetic radiation signal, and marking space-time coordinates; performing deep processing on the acquired signals, separating multi-frequency signals, correcting attenuation, and constructing a multi-dimensional feature vector; based on the feature vector, a space-time abnormal mode is extracted to position a damage point, through damage type classification, region area measurement and depth evaluation, the damage severity is quantified, accurate data support is provided for pipeline maintenance, and the accuracy and reliability of metal pipeline anticorrosive coating detection are effectively improved.

The invention provides a noise reduction method and device, equipment and a medium, weighting processing is performed on a first time domain reference noise signal vector to obtain a second time domain reference noise signal vector, a sampling point closer to the current moment is endowed with a higher weight, the noise reduction performance is optimized, and the noise reduction performance is improved by updating a weight coefficient and the second time domain reference noise signal vector. Therefore, the noise cancellation signal is determined on the time domain, filtering on the time domain is realized, the noise cancellation signal is generated to carry out noise reduction processing on the primary noise in the target noise reduction area, extra time delay caused by frequency domain filtering is avoided, noise reduction time delay is reduced, and the filtering effect is improved. The updating process of the weight coefficient is completed in the frequency domain, updating is achieved in the frequency domain with smaller blocks, the calculation amount is reduced, the interval time of noise reduction processing is shortened, the calculation efficiency of the algorithm is improved, the processing delay is reduced, and the noise reduction effect is improved.

The invention provides an active noise reduction method capable of restraining divergence and a microphone vibration reduction device, belongs to the technical field of automobile active noise reduction, and detects the sealing condition of a cab through automobile door sensors, automobile window sensors and skylight sensors. If the microphone is closed, noise signals are collected by using errors on the microphone vibration damping device and a reference microphone, and secondary noise signals are generated by an active noise reduction filter and are played by a loudspeaker; if yes, the engine rotating speed in the vehicle CAN signal is read, and a reference noise signal is generated and played. A cab sealing state and an active noise reduction system operation state are displayed on an instrument panel, and the operation state is determined according to noise signal acquisition and noise reduction signal playing conditions. The noise reduction mode can be automatically switched according to the cab sealing state, the noise in the vehicle is effectively reduced when the cab is sealed, the reference noise is reasonably played when the cab is opened, and the influence on the robustness of a system filter is reduced. Related information is displayed on the instrument panel, so that a user can visually know the cab state and the system operation condition.

The invention discloses a multi-mode composite fault diagnosis method and system for a variable-frequency asynchronous motor. The method comprises the following steps: synchronously acquiring a vibration acceleration signal X, a rotating speed signal R and a current signal D of the variable-frequency asynchronous motor; performing adaptive noise reduction on the vibration acceleration signal X by using a rotating speed driven harmonic wavelet packet-variational mode decomposition method to obtain a reconstructed signal Y; filtering the current signal D through dual-mode collaborative filtering and soft reference noise, and outputting a noise reduction signal S through dual-channel fusion processing; extracting a pass frequency speed effective value and an acceleration kurtosis from the reconstructed signal Y, and extracting a current effective value, a peak coefficient, a degradation coefficient and a harmonic distortion sequence from the noise reduction signal S; vibration, rotating speed and current characteristics are fused to set expert mechanism model rules, and fault types are output according to the rules. The early weak fault sensitivity can be improved, and accurate, real-time and automatic early warning of composite faults such as bearing damage, rotor bar breakage and air gap eccentricity under complex working conditions can be realized without manual intervention.

The invention discloses a video pulse wave extraction method and system based on automatic noise recognition. The method comprises the steps that a face area is divided into a plurality of sub-areas; calculating the signal-to-noise ratio of each sub-region in the pulse frequency band, and performing weighted fusion on the original pulse signals of each sub-region according to the signal-to-noise ratio; performing ensemble empirical mode decomposition on the fusion signal; according to the peak-to-peak interval variation coefficient, the amplitude stability index and the spectrum purity, determining a noise component dominated by the motion artifact; performing adaptive filtering processing on the fused signal by using an adaptive filter and taking a reference noise signal as input; carrying out distortion detection on the filtered fusion signal, and carrying out signal restoration processing on a detected distortion region to obtain an extracted pulse signal; in order to solve the problem that in the prior art, when video pulse signals are extracted, motion artifacts and physiological signal frequency spectrums are overlapped, and consequently signal separation of a traditional frequency domain filtering method fails, effective signal separation and accurate pulse signal extraction are achieved.

The invention discloses a low-noise low-power-consumption cascade input comparator which comprises a pre-amplification stage and a latch stage, the pre-amplification stage is used for amplifying a pair of differential signals input from the outside, the amplified differential signals are obtained and output to the latch stage, and the latch stage is used for latching the amplified differential signals. The latch stage is used for keeping a latch state and outputting a differential comparison result, the pre-amplification stage is used for inputting a pair of differential signals by arranging a three-stage cascade structure consisting of three pairs of input geminate transistors, and is used for preventing all connection nodes of the pre-amplification stage and the latch stage from being completely discharged to the ground by arranging a pair of cross coupling geminate transistors, so that charge storage is realized; the circuit has the advantages of low input reference noise, no consumption of a large amount of extra area and power consumption, small area and low power consumption.

The present application relates to the technical field of compression spring resonance frequency detection, and specifically relates to a compression spring resonance frequency detection method and system, the method comprising the following steps: collecting a mixed vibration signal of a compression spring to be measured; collecting a background noise signal of a test device as a reference noise signal; collecting a synchronous signal of a specific noise source of the test device; performing adaptive noise cancellation processing on the mixed vibration signal, the adaptive noise cancellation processing performing initial parameter training using the reference noise signal and using the noise source synchronous signal as a real-time reference input to separate a pure vibration response signal of the compression spring to be measured; performing frequency spectrum analysis on the pure vibration response signal; identifying a wideband vibration response region of the compression spring to be measured, the wideband vibration response region having a continuous amplitude higher than a preset energy threshold and having a specific width; calculating a characteristic parameter of the wideband vibration response region; and judging a degradation state of the compression spring to be measured according to the characteristic parameter. The above improves the accuracy of degradation judgment.

The invention relates to a metal pipeline corrosion nondestructive monitoring method, device and equipment and a medium. The method comprises the following steps: acquiring high-frequency eddy current data, low-frequency eddy current data and reference noise data of a pipeline; based on the reference noise data, performing adaptive noise removal and modal decomposition on the high-frequency eddy current data and the low-frequency eddy current data to obtain an intrinsic corrosion characteristic signal; the intrinsic corrosion characteristic signal comprises a high-frequency intrinsic corrosion characteristic signal and a low-frequency intrinsic corrosion characteristic signal; performing multi-band feature extraction on the intrinsic corrosion feature signal to obtain a fusion feature vector; and carrying out inversion on the fused feature vector to obtain corrosion depth distribution data. By adopting the method, the corrosion characteristics of the deep and shallow layers of the non-ferromagnetic metal pipeline can be monitored at the same time, and the false detection rate of the deep corrosion condition in a strong interference environment is reduced.

The application provides a high-density surface electromyography signal hardware filtering device and method based on FPGA, comprising: an sEMG signal input module for collecting multi-channel noisy surface electromyography signals and at least one reference noise signal; an LMS adaptive filtering calculation module for receiving the noisy surface electromyography signals and the reference noise signal, adopting a fixed-point number operation system and executing an LMS adaptive filtering algorithm through a parallel pipeline structure to output a filtered error signal; a data buffer and output interface module for buffering and outputting the filtered surface electromyography signals; and an FPGA main control module connected with the sEMG signal input module, the LMS adaptive filtering calculation module and the data buffer and output interface module for controlling the work of each module. The application realizes the LMS adaptive filtering algorithm on the FPGA platform, and solves the problems of insufficient real-time performance and large processing delay of software filtering when facing high-density multi-channel sEMG signal processing.

The invention relates to the technical field of intelligent sensing systems, and discloses a method and a device for detecting the bearing capacity of a pavement engineering construction material, and the method comprises the steps: synchronously collecting a mixed signal and a pure noise signal through a main measurement sensor and a reference noise sensor, constructing a multi-dimensional observation matrix, and carrying out the centralization and whitening preprocessing, thereby obtaining the bearing capacity of the pavement engineering construction material. And blind source separation is carried out by adopting an independent component analysis algorithm based on maximized non-Gaussian property, a target bearing capacity response signal with a super Gaussian characteristic is identified according to a kurtosis value of each source signal component, and then a dynamic deformation modulus is calculated. The system comprises a sensor array module, a synchronous acquisition module, a data processing module based on FPGA hardware acceleration and a result output storage module. According to the invention, strong coherent vibration noise can be effectively suppressed, signal details are completely reserved, and high-precision and real-time bearing capacity detection is realized.

The application relates to a vehicle sealing performance detection method and device, a vehicle, a storage medium and a product. The method comprises the following steps: acquiring original acoustic signals collected by a main microphone array arranged in a to-be-detected area of a vehicle and reference noise signals collected by a reference microphone array arranged in a reference area of the vehicle; performing background noise filtering processing on the original acoustic signals based on the reference noise signals to obtain target acoustic signals; extracting target acoustic features corresponding to the target acoustic signals; determining a feature similarity between the target acoustic features and reference acoustic features corresponding to a current running state of the vehicle; and in a case where the feature similarity is smaller than a similarity threshold, determining a sealing performance detection result of the to-be-detected area according to the target acoustic features. The method can continuously and comprehensively detect the sealing performance of the vehicle during driving.

The embodiment of the invention discloses a model training method and device, a multimedia resource generation method and device and computer equipment. The method comprises the steps that multimedia features and text features are acquired; in the first training stage, training parameters are adopted, and a to-be-trained multimedia generation model is trained at least once based on the text features, the first noise-added multimedia feature set and the first reference noise feature set, so that a pre-trained multimedia generation model is obtained; and in a second training stage, training the pre-trained multimedia generation model for multiple times by adopting the compressed training parameters based on the text features, the second noise-added multimedia feature set and the second reference noise feature set to obtain a trained multimedia generation model. In this way, the effect of generating multimedia through the multimedia generation model can be improved.

The invention provides a multi-channel synchronous sampling system for weak current detection and a control method thereof. According to the method, a low-noise trans-impedance amplifier array is used for converting multiple paths of weak current signals into multiple paths of analog voltage signals; at least one reference channel for outputting reference noise data; the synchronous analog-to-digital conversion array is used for synchronously sampling the multi-path analog voltage signals and the reference noise data after receiving the global conversion starting signal sent by the field programmable gate array, and converting the multi-path analog voltage signals and the reference noise data into multi-path synchronous first digital data to be output; the field programmable gate array is used for receiving the first digital data and carrying out delay compensation and amplitude correction on the first digital data to obtain multiple paths of time-aligned and amplitude-normalized second digital data; and a filter for referencing the first digital data converted from the reference noise data and eliminating the common-mode noise component in the second digital data to output purified third digital data.

The present application provides a signal processing method and a related apparatus. The method comprises: acquiring N quantized audio features of raw non-speech audio, wherein N is a positive integer greater than or equal to 1; determining a first predicted silence audio feature based on the N quantized audio features and reference noise; and determining, based on the first predicted silence audio feature and a first neural network, first non-speech audio corresponding to the raw non-speech audio.

This invention provides an active noise cancellation method and microphone vibration damping device that suppresses divergence, belonging to the field of automotive active noise cancellation technology. It detects the cabin's sealing status using sensors on the doors, windows, and sunroof. If sealed, noise signals are collected using the error on the microphone vibration damping device and a reference microphone. These signals are then processed by an active noise cancellation filter to generate a secondary noise signal, which is played through a speaker. If open, the engine speed is read from the vehicle's CAN signal to generate a reference noise signal, which is also played. The dashboard displays the cabin sealing status and the active noise cancellation system's operating status, determined based on the noise signal collection and noise cancellation signal playback. This invention automatically switches noise cancellation modes according to the cabin's sealing status, effectively reducing interior noise when sealed and appropriately playing reference noise when open, minimizing the impact on the system filter's robustness. Relevant information is displayed on the dashboard, allowing users to intuitively understand the cabin status and system operation.

The invention relates to a pressure sensor and a sensor. The pressure sensor comprises a sensor element (WB) in the form of a Wheatstone bridge and having four piezoresistive resistors (R1, R2, R3, R4), and a reference element (RW) in the form of a reference Wheatstone bridge and having four piezoresistive reference resistors (R5, R6, R7, R8), the reference resistors (R5, R6, R7, R8) being arranged in the same way as the resistors (R1, R2, R3, R4), the pressure sensor being arranged on a single crystal and comprising a cavity closed on one side by a diaphragm, the resistors (R1, R2, R3, R4) being arranged at least partially on the diaphragm, the reference resistors (R5, R6, R7, R8) not being arranged on the diaphragm, the first to fourth resistors (R1, R2, R3, R4) respectively being identical to the fifth to eighth resistors (R5, R6, R7, R8) in that they are constructed in the same way, and the reference Wheatstone bridge (RW) serving as a reference noise source for subsequent signal processing.

The invention relates to a signal detection method based on actual noise weak correlation, and the method comprises the following steps: S1, setting a sampling rate through a Nyquist sampling theorem according to the bandwidth of a to-be-detected signal; s2, collecting a noise sample and a to-be-tested signal sample according to a set sampling rate; s3, calculating a reference noise signal according to the collected noise sample; s4, setting a delay interval, and calculating a cross-correlation distance between the reference noise signal and the to-be-measured signal sample according to the delay interval; step S5, judging whether the detection threshold can realize detection according to the cross-correlation distance, if so, storing algorithm parameters, and otherwise, returning to the step S1; and step S6, sampling a to-be-detected signal according to the sampling rate stored in the step S5, and performing signal detection on a to-be-detected signal sequence obtained by sampling by using the stored algorithm parameters. According to the invention, the signal detection is completed by using the related information between the signal and the noise, and the probability of signal detection under a low signal-to-noise ratio is improved.

The application discloses a continuous monitoring single-arm multi-modal acquisition system and a feature fidelity denoising method, and relates to the physiological signal monitoring technical field.The system comprises an electrocardiosignal acquisition module, a multi-modal signal acquisition module, a communication module, a power module and an upper computer.The method is applied to the system and specifically comprises the following steps: taking a motion signal as a reference input, modeling and adaptively suppressing motion noise in an electrocardiosignal to obtain a signal after preliminary denoising; and inputting the signal after preliminary denoising and reference noise information into a pre-trained reconstruction model to generate a reconstructed signal.The application overcomes the technical problem that it is difficult to balance portability and signal acquisition quality in an existing off-site scene.The application can be widely applied to the physiological signal monitoring technical field.

The invention discloses a receiving sensitivity testing method and system based on a Bluetooth HCI command and conduction isolation, and relates to the technical field of wireless communication testing. The method comprises the following steps: constructing an electromagnetic shielding test environment, and carrying out system calibration in the test environment to obtain a power calibration value and an inherent loss value; measuring background noise power of each channel, and determining an optimal test channel and a power compensation factor in combination with a reference noise threshold; according to the Bluetooth communication protocol physical layer parameters and the target packet error rate, determining the number of theoretical transmitted data packets, and calculating a judgment threshold of the number of successfully received data packets; searching a critical attenuation value, iteratively adjusting the attenuation value of the programmable attenuator according to a judgment threshold value, and adjusting an attenuation step value according to the change rate of the number of successfully received data packets; and calculating the final receiving sensitivity based on the power calibration value, the inherent loss value, the critical attenuation value and the power compensation factor. By optimizing electromagnetic shielding, channel selection and three-stage scanning, the accuracy of receiving sensitivity testing is improved.

This application relates to a method, apparatus, and computer device for cross-domain adversarial example detection. The method includes: acquiring an unknown sample set and the signal-to-noise ratio (SNR) of the transmission channel for each unknown sample in the unknown sample set; filtering adversarial examples and samples to be detected from the unknown sample set according to a preset reference noise distribution; inputting the samples to be detected into multiple pre-trained adversarial detection models to obtain multiple adversarial probabilities for the samples to be detected; the adversarial detection models are used to predict the probability that the sample to be detected is an adversarial example, and different adversarial detection models target different types of interference; and obtaining the adversarial detection result of the sample to be detected based on the SNR of the transmission channel of the sample to be detected and the adversarial probabilities of the sample to be detected. This method can improve the accuracy of adversarial example detection.

This specification provides one or more embodiments of an in-vehicle audio equalization method, apparatus, storage medium, and program product. The method includes: acquiring a background noise signal of the current in-vehicle environment and determining the gain difference spectrum of the background noise signal relative to a reference noise signal of a quiet environment; performing track-by-track processing on the currently playing media audio signal to obtain multiple independent audio tracks corresponding to different frequency ranges; for each independent audio track, determining a corresponding gain adjustment amount from the gain difference spectrum based on its frequency range, and applying a corresponding gain adjustment to the independent audio track when the gain adjustment amount meets a gain condition; merging the signals of each independent audio track to output as a processed media audio signal, thereby improving speech clarity and music listening experience while significantly improving the accuracy and naturalness of audio equalization.

The invention relates to the technical field of deep groove bearing detection, in particular to a deep groove bearing rolling body surface defect detection method and system. Through layered vibration signal acquisition, low-frequency reference noise self-adaptive elimination and rolling body motion trail modeling, accurate detection of micron-sized defects on the surface of a deep groove bearing rolling body is realized, different signal channels of the rolling body and a bearing seat are effectively covered by constructing a multi-source vibration sensing layered acquisition system, and the detection precision of the deep groove bearing rolling body is improved. By synchronously acquiring low-frequency reference noise and high-frequency micro-impact signals and performing self-adaptive time domain noise cancellation, interference such as equipment background noise and gear meshing is remarkably suppressed, a rolling body motion and impact pulse correlation model is established by using structural parameters and real-time rotation parameters, and pure micro-impact signals are directionally enhanced, so that the precision of the rolling body motion and impact pulse correlation model is improved. Through multi-dimensional feature extraction and three-dimensional closed-loop analysis, accurate positioning and authenticity verification of defect features are realized, and graded early warning of rolling body surface defects is completed in combination with a defect grade threshold and an early warning rule.

The invention discloses a pulse signal wireless transmission and calibration method based on cloud-edge cooperation, and relates to the technical field of signal transmission, and the method comprises the steps: an edge node collects reference noise in a transmission environment, generates estimated noise through an adaptive filter, carries out the denoising of a received pulse signal with noise, and carries out the amplitude distortion detection and preliminary correction; edge nodes exchange timestamps through a self-organizing network to realize initial time synchronization, and a cloud absolute time signal is combined, a local clock is finely adjusted by adopting a PI control algorithm, and time unification of the whole network is realized; the edge nodes detect the actual feature points of the pulse signals through a sliding window, align the actual feature points with the standard feature points, adjust window parameters, and perform local calibration in combination with local initial calibration parameters; the cloud end issues initial calibration parameters to the edge nodes, the edge nodes carry out secondary calibration and upload local optimization parameters, the cloud end carries out weighted fusion to generate global optimal parameters, iterative optimization is carried out until calibration errors are stable, and the real-time performance and accuracy of signals are remarkably improved.

The present disclosure provides a speech synthesis model training method, a speech synthesis method, an apparatus and a device, belonging to the technical field of computer. The first speech synthesis model comprises a language network and a diffusion network; the method comprises: processing a sample text and a sample audio through the language network to obtain a first audio feature, the sample text and the sample audio are consistent in content, and the first audio feature comprises a content feature of the sample text and an acoustic feature of the sample audio; performing noise prediction on the sample audio added with reference noise based on the first audio feature through the diffusion network to obtain a predicted noise; determining a first training loss based on the reference noise and the predicted noise; and updating parameters of the language network and the diffusion network in the first speech synthesis model based on the first training loss to obtain a second speech synthesis model. The above technical solution achieves the purpose of end-to-end training of the speech synthesis model, simplifies the training steps, and is beneficial to improving the training efficiency of the model.

The application provides a lower limb movement state electroencephalogram movement artifact removal method, equipment and a storage device, the corresponding electromyogram reference noise is estimated according to the movement speed and speed change of the lower limb; the movement artifact is preliminarily eliminated through adaptive noise cancellation to the electromyogram reference noise; the signal processed through the adaptive noise cancellation is subjected to variational mode decomposition, and the intrinsic mode function of each order variable is obtained; the intrinsic mode function of each order is subjected to independent component analysis, and the independent component is obtained; the time domain feature, the frequency spectrum feature and the intersequence similarity of the independent component are calculated, then the artifact is identified and removed through hierarchical clustering identification, and the remaining components are reconstructed to obtain the processed independent component, i.e. the reconstructed electroencephalogram signal. The application has the beneficial effect that the interference caused by electromyogram noise when the lower limb speed changes can be quickly responded.

The present application relates to the field of signal denoising, and provides a PPG signal light-weight diffusion denoising method, system, device and medium, which comprises generating reference noise and obtaining a pure signal to obtain a reference signal; the pure signal is added with noise to obtain a diffusion noise signal; a multi-scale one-dimensional convolution structure is constructed, the diffusion noise signal and the reference signal are respectively convolved through the multi-scale one-dimensional convolution structure to obtain a double-branch feature sequence; light-weight attention calculation is performed on the double-branch feature sequence to obtain an updated double-branch feature sequence, multi-layer residual fusion and noise prediction are performed on the updated double-branch feature sequence to obtain a noise prediction result; the diffusion noise prediction loss, the average spectral amplitude difference and the waveform form loss of the noise prediction result are calculated to obtain a prediction loss, and the target prediction model is obtained through training according to the prediction loss, and the PPG signal is denoised through the target prediction model.

The invention discloses a signal noise reduction and power supply optimization method for an earphone internal circuit, and the method comprises the core steps: collecting a real-time power supply voltage, a real-time interference signal frequency, a real-time circuit temperature and a real-time load resistance based on a synchronous collection time sequence control logic, correcting a reference noise reduction coefficient based on a parameter coupling effect, and obtaining a dynamic noise reduction coefficient; and adjusting the reference power supply current based on the dynamic noise reduction coefficient and the real-time load resistance to obtain a self-adaptive power supply current, and calculating a noise reduction-power supply comprehensive performance index based on the dynamic noise reduction coefficient and the self-adaptive power supply efficiency. According to the method, through synchronous acquisition and multi-parameter coupling linkage adjustment, the core problem of independent optimization of noise reduction and power supply in the prior art is solved, collaborative adaptation of noise reduction and power supply is realized, and the operation stability of an earphone circuit is improved.

The application provides a speech enhancement method based on BGRU and reference noise, and contains the following network structure: during reasoning, two-way input is accepted: reference noise and noisy speech, the two-way noise is respectively input to an enhancement network module through a short-time Fourier transform encoder to obtain noise estimation and speech estimation, and the speech estimation is input to an inverse short-time Fourier transform to obtain the speech after noise reduction; during training, four-way input is accepted: reference noise, noisy speech, noise true value and speech true value, the four-way noise is respectively input to a short-time Fourier transform encoder, the encoded reference noise and noisy speech are calculated through the enhancement network module to obtain noise estimation and speech estimation; the encoded speech true value and noise true value, and the noise estimation and speech estimation are all input to a 3CL loss module, the loss is calculated and fed back to the enhancement network module to adjust the parameters and improve the performance of the enhancement network module. The application can improve the speech quality and intelligibility under the condition of extremely low signal-to-noise ratio.

This invention relates to the field of audio processing technology, and more particularly to an audio optimization method, apparatus, computer device, and medium based on artificial intelligence. The method uses a linear layer to map the spectral envelope of the audio to be optimized to obtain envelope features; uses an embedding layer to embed standard audio parameters as parametric features; uses a prediction model to predict the fusion features of the envelope features and parametric features to obtain a predicted pitch curve; uses a noise-adding model to add noise to the Mel spectrum of the audio to be optimized to obtain a noise-adding result; uses a noise estimation model to calculate the noise of the noise-adding result to obtain predicted noise; updates the noise estimation model based on the predicted noise, the actual noise, and the predicted pitch curve; uses the updated noise estimation model to calculate the reference noise of the noise-adding result; and denoises the noise-adding result based on the reference noise to obtain the optimized Mel spectrum. Combining pitch information with the optimized noise estimation model ensures that the denoising process meets pitch requirements, thus improving the audio optimization effect.