12 results about "Noise weighting" patented technology

The invention provides an engine active noise reduction method and system and a vehicle, and the method comprises the steps: S1, determining the noise weight of each noise reduction region based on the personnel distribution in each noise reduction region, and weighting the noise collection value in the corresponding noise region based on the noise weight to form weighted noise; and S2, inputting the weighted noise into a control filter, generating a sound wave signal for reducing the noise in the noise reduction area by the control filter, and transmitting the sound wave signal to each noise reduction area. According to the change of people in the noise reduction area in the vehicle, the noise reduction effects of different noise reduction areas are adjusted in a self-adaptive mode, the noise reduction effect of the noise reduction area where people exist is mainly improved, and the noise reduction effect is more intelligent; the method is suitable for various types of vehicles including cars, SUVs, commercial vehicles and the like, and has high universality and market adaptability.

The invention relates to the technical field of wearable device mutual-seeking positioning, in particular to a wearable device mutual-seeking positioning method and system based on an entry and exit principle, and the method comprises the steps: setting a base station, obtaining angle and distance information through a multi-antenna array, and carrying out the time slot positioning; constructing a shielding model by utilizing the equipment attitude information, and identifying a sight distance communication and signal shielding blind area; the Kalman filtering noise weight is adjusted by adopting a dynamic sector confidence gating strategy, a relative pose vector is solved in combination with geometric constraints, and a guide instruction is generated; in the near-field verification stage, a coding vibration sequence is transmitted, time domain wave packet cross correlation is calculated to confirm physical contact, and automatic search quitting is achieved. According to the invention, through a shielding model and a dynamic gating strategy, positioning drift caused by shielding is effectively inhibited, and the guiding robustness in a complex environment is improved; by means of coding vibration and a cross-correlation verification mechanism, misjudgment and real contact of the partition wall are accurately distinguished, and intelligent closed loop of guiding while entering and returning while contacting is achieved.

The invention discloses an interference suppression method, system and device based on frequency point quality cognition and a medium, and relates to the field of frequency hopping communication. A transmitting end adopts a frame structure of'synchronous head sequence + information sequence + synchronous tail sequence ', and forms a frequency hopping pulse signal for radiation after processes of coding, framing and frequency hopping modulation; a receiving end processes a received frequency hopping pulse signal after signal detection and timing synchronization, calculates a noise power value of each frequency hopping pulse after frequency hopping pulse synchronization head and tail sequences are correlated with a local sequence, calculates a noise weight value of each pulse, multiplies a weight factor with a pulse signal after channel equalization, and obtains a frequency hopping pulse signal. By reducing the signal observed quantity weight of the interfered frequency point, the influence of the confidence coefficient reduction caused by the interference signal on the channel decoding of the whole communication system is suppressed, so that the reliable transmission under the condition of low signal-to-noise ratio and strong interference is realized.

The invention discloses a sparse Bayesian out-of-grid orientation estimation method under non-Gaussian noise, which comprises the following steps of: constructing a hierarchical noise prior model based on Student-t distribution, a double-layer sparse signal prior model based on a binary indication variable and the like, and inferring approximate posteriori distribution of iteratively updating parameters by using Bayesian variation; the noise precision weight is adaptively adjusted according to the reconstructed residual energy so as to suppress impulse noise, and random pseudo peaks are effectively removed through a binary indication variable judgment mechanism. And finally, adaptively correcting the grid offset by using the residual gradient information subjected to anti-noise weighting so as to eliminate the off-grid effect. Compared with the traditional method, the method provided by the invention not only can carry out DOA estimation under non-Gaussian noise, but also has higher DOA estimation precision. The method disclosed by the invention still has relatively high performance under the conditions of few snapshots and low signal-to-noise ratio, so that the method has relatively high application value in actual engineering.

The invention discloses a Beidou-based high-precision differential positioning adjustment system and method, and relates to the technical field of satellite navigation and high-precision localization. Time-frequency analysis is carried out on a multi-frequency carrier phase observation sequence, signal-to-noise ratio monitoring is combined, and a flicker identifier and a noise weight are constructed; the system can identify frequency points and observation sections which are seriously polluted by flicker at an epoch level, and the weight of the system is actively reduced at a combination stage; secondly, under the constraint of maintaining ambiguity integer characteristics, the coefficient combination takes equivalent noise variance and residual ionosphere high-order term error sensitivity as optimization targets, and considers the observation noise level and the suppression capability for high-order ionosphere residual errors, so that on one hand, the reliability of carrier phase integer ambiguity calculation is maintained, and on the other hand, the reliability of carrier phase integer ambiguity calculation is improved; and on the other hand, the sensitivity degree of combined observation to high-order ionosphere residual errors is reduced, so that high-precision differential positioning of a low-latitude region has higher robustness.

This invention discloses an intelligent drill bit attitude segmented adaptive filtering method, system, and storage medium. The intelligent drill bit attitude segmented adaptive filtering method includes: acquiring multi-source measurement signals from the drill bit and performing preprocessing and dynamic-static separation; extracting features to construct discrimination indices, dividing the working conditions into drilling start-up impact, continuous cutting, and idling separation; calculating acceleration reliability coefficients; establishing a full-state filtering model including attitude and zero bias for attitude prediction and temperature drift compensation; matching measurement update strategies according to identified working conditions; adaptively adjusting the noise covariance matrix based on the reliability coefficients; performing smoothing processing on the preliminary results; and outputting the final attitude parameters. This invention can dynamically switch filtering strategies and adaptively adjust noise weights according to changing working conditions, isolate strong impact interference, overcome zero bias drift caused by high temperatures, and improve the accuracy and robustness of attitude calculation in complex drilling environments.

The invention relates to a feed-forward equalizer performance optimization method. The method comprises the following steps: S1, acquiring an input signal; s2, inputting the input signal into a feed-forward equalizer, and obtaining an output signal of the feed-forward equalizer; s3, the output signals are processed, noise signals are obtained and grouped, each group of noise signals are weighted and summed, then symbols are taken, and the symbols of the noise weighted sums are obtained; and S4, judging whether a symbol of the noise weighted sum needs to be corrected or not, if the symbol needs to be corrected, correcting an output signal of the feed-forward equalizer according to a correction coefficient and then outputting the corrected output signal, if the symbol does not need to be corrected, outputting the output signal of the feed-forward equalizer, returning the corrected signal or the uncorrected signal as a next-level output signal to the step S3, and if the symbol does not need to be corrected, returning the corrected signal or the uncorrected signal to the step S3. And S4 is repeated until the preset number of iterations is reached. According to the invention, the performance of the FFE which is low in complexity and widely applied is improved.

The present invention relates to a laser point cloud registration method and system for autonomous driving. The method comprises the following steps: obtaining a source point cloud and a target point cloud captured by a laser radar in the autonomous driving, initializing noise weights based on density, and calculating fast point feature histograms of the source point cloud and the target point cloud; downsampling the source point cloud and the target point cloud according to the fast point feature histogram and the number of iterations, calculating the matching probability of source points in the source point cloud to target points in the target point cloud, counting source points with matching probabilities lower than a threshold as noise points, and calculating the noise weights for the next iteration using a dynamic noise estimation method based on the number of noise points; optimizing a transformation matrix and a translation vector based on the matching probability, determining the change in the transformation matrix and the translation vector relative to the last calculated transformation matrix and the translation vector, and stopping updating the transformation matrix and the translation vector when the change meets a preset condition; and transforming the source point cloud using the transformation matrix and the translation vector to complete the registration of the source point cloud and the target point cloud.

The present invention belongs to the technical field of infrared camera imaging, and discloses an infrared camera active noise reduction system, a control method and an application thereof. The system is provided with a data acquisition device, a fixed pattern noise suppression device, an active sampling control device, and a training and decision-making device. The data acquisition device is respectively connected to the fixed pattern noise suppression device and the active sampling control device through connection lines. The fixed pattern noise suppression device is connected to the active sampling control device through a connection line. The data acquisition device is provided with an AD converter, a uniform radiation source, a servo mechanism, an optical unit and a detector. The training and decision-making device is provided with a single spectral band noise statistic, a data accumulation processor, a noise weighting statistic, and a threshold judge. The present invention has low comprehensive noise. At the same time, since the sampling points of each channel or spectral band can be adjusted, the instantaneous ground bounce disturbance caused by simultaneous sampling between channels / spectral bands is also suppressed, and it has the characteristics of strong real-time performance and small data operation volume.

The present application relates to wearable device mutual search positioning technical field, specifically wearable device mutual search positioning method and system based on approach and retreat principle, including setting base station utilizes multi antenna array to obtain angle and distance information, time slot positioning is carried out simultaneously;Utilize device posture information to construct shielding model, identify line-of-sight communication and signal shielding blind area;Adopt dynamic sector confidence gating strategy to adjust Kalman filter noise weight, combine geometric constraint to solve relative pose vector and generate guide instruction;In near field verification stage, transmit coded vibration sequence, calculate time domain wave packet cross correlation to confirm physical contact, realize automatic exit search.The present application effectively suppresses the positioning drift caused by shielding through shielding model and dynamic gating strategy, improves the guidance robustness in complex environment;Utilize coded vibration and cross correlation verification mechanism, accurately distinguish partition wall misjudgment and real contact, realize intelligent closed loop of approach and guidance, contact and retreat.

The invention belongs to the technical field of power load prediction, and discloses a power load prediction model training method and a related device. The power load prediction model training method comprises the following steps: performing signal decomposition on a historical power load sequence by using ICEEMDAN to obtain a plurality of subsequences; during signal decomposition, the noise weight and the decomposition frequency of the ICEEMDAN adopt an optimal parameter combination (Nstd *, NE *) found by the IHHO; splicing the plurality of obtained subsequences and the plurality of selected exogenous variables in sequence in a time dimension to construct a training sample; and the iTransform model is trained, and a trained power load prediction model is obtained. According to the technical scheme, the technical problems that in the prior art, ICEEMDAN parameters are set depending on experience, the adaptive capacity is insufficient, the prediction model structure is complex, the training cost is high, the decomposition and modeling process lacks a collaborative optimization mechanism, and the external feature interference robustness is poor are solved.

The invention provides an unmanned aerial vehicle acoustic detection method and system based on MFCC feature improvement, and the method comprises the steps: adding the sound pressure level sequence analysis of an unmanned aerial vehicle acoustic signal on the basis of a conventional MFCC feature, thereby determining a new frequency transformation interval, designing a corresponding Mel frequency transformation formula, obtaining an improved MFCC feature, and improving the detection precision. Inputting the improved MFCC features into an i-vector extraction model for training to obtain an i-vector of an unmanned aerial vehicle acoustic signal, training a noise compensation model and a noise weight model by using the i-vector, inputting a to-be-detected unmanned aerial vehicle acoustic signal into the models to obtain an i-vector of the to-be-detected unmanned aerial vehicle acoustic signal with background noise removed, and matching the i-vector with the i-vector in a matching library to obtain a matching result of the to-be-detected unmanned aerial vehicle acoustic signal. And obtaining a corresponding unmanned aerial vehicle model. According to the method, the problem that the traditional MFCC feature representation capability is insufficient in a complex scene is solved, and the accuracy of the unmanned aerial vehicle classification task is effectively improved.