14results about How to "Solve noise interference" patented technology

The application relates to the technical field of wind turbine fault detection, and provides a wind turbine transmission chain fault detection method and system, which comprises the following steps: acquiring multi-dimensional monitoring data of a wind turbine transmission chain; performing standardized preprocessing on the multi-dimensional monitoring data based on operation condition characteristics and signal transmission characteristics to obtain effective monitoring data; performing feature extraction on the effective monitoring data based on time-frequency feature analysis and fault feature enhancement processing to obtain transmission chain full-link fault features; combining historical operation data and real-time operation trends through data statistics and time sequence analysis algorithms to construct and update adaptive detection thresholds; and comparing the full-link fault features with the adaptive detection thresholds based on multi-index fusion decision and transmission chain fault conduction mechanism analysis to obtain fault detection results. The application solves the problems of multi-source data asynchronization and weak fault identification difficulty, and realizes high-precision fault detection and accurate positioning under variable conditions.

The invention belongs to the technical field of signal processing and deep learning, and particularly relates to a pressure signal reconstruction method based on deep learning, which comprises the following steps: classifying an input dynamic pressure signal, and judging whether the signal is complete or not; for an incomplete signal, missing data completion is carried out by using a time-frequency feature fusion filling method based on LSTM and GAIN, and then noise filtering is carried out by using a noise reduction method based on CEEMDAN-wavelet packet fusion LSTM; for a complete signal, a noise reduction method based on CEEMDAN-wavelet packet fusion LSTM is directly applied to remove noise interference; and outputting the reconstructed complete and clear dynamic pressure signal. The method can effectively solve the problems of signal loss and noise interference in an extreme environment, provides reliable technical support for research and development, testing and performance evaluation of industrial equipment, and has important industrial application value and wide market prospect.

The application discloses a kind of multi-modal sentiment analysis method and device based on feature decoupling and guiding correction, it is related to sentiment analysis technical field, including: build initial multi-modal sentiment analysis network, initial multi-modal sentiment analysis network includes multi-modal feature extraction mapping block, cross-modal interaction enhancement coding block, feature decoupling block, guiding formula gate semantic correction block and low rank fusion sentiment analysis block, determine the total loss value of training sentiment analysis result of training multi-modal sentiment data, according to total loss value to initial multi-modal sentiment analysis network is optimized iteration, until determining target multi-modal sentiment analysis network, using target multi-modal sentiment analysis network exports the target sentiment analysis result of the multi-modal sentiment data to be analyzed.Based on the above scheme, help to improve multi-modal fusion sentiment recognition precision.

This invention provides a method for constructing and detecting remote photoplethysmography (PPG) signals and heart rate. The construction method includes: acquiring and preprocessing a sequence of facial video images as an initial dataset; acquiring PPG signals as a label set; constructing a PPG signal and heart rate detection model, using the initial dataset as input and the label set as output to train the PPG signal and heart rate detection model, and the model outputting the final remote PPG signal. This invention adds a multi-level feature fusion module and a multi-scale enhancement module to the encoder and decoder, respectively, to ensure that effective physiological features highly correlated with the temporal information of the remote PPG signal are retained during feature extraction, thereby recovering accurate remote PPG signals and improving model performance. The signal estimation module performs filtering operations on the predicted signal, which can better adapt to waveform distortion values ​​to a certain extent and improve the model's prediction accuracy.

The present application relates to the technical field of enterprise management and risk monitoring, and specifically discloses a method and system for penetrating risk monitoring of invested enterprises based on supply chain conduction, which accurately reflects the dynamic changes of the supply relationship between enterprises by constructing a supply chain meta-network containing dynamic conduction impedance tags; calculates the propagation tendency of risks in the supply chain using a risk gravity model, dynamically generates a risk propagation chain pointing to the invested enterprise, and realizes forward-looking early warning of risks; focuses on high-risk conduction paths to improve monitoring efficiency and accuracy, and solves the problem of analysis delay caused by massive data; at the same time, the system has closed-loop adaptive learning capability, and updates the supply chain meta-network parameters in reverse according to the actual risk conduction results, continuously improving the accuracy of the risk assessment model. The present application can identify potential risks in advance, provide timely and accurate risk early warning for investment institutions, and effectively improve the initiative of risk management.

The utility model discloses a kind of mute magnetic shaft and keyboard, wherein, a kind of mute magnetic shaft includes shell, button piece, magnet piece and spring, shell includes the upper cover and base connected, the upper cover is equipped with avoiding hole, button piece is slidably worn in avoiding hole, button piece is used to be connected with keycap, button piece and upper cover at least one is equipped with first buffer, magnet piece is located in button piece, magnet piece and base at least one is equipped with second buffer, spring is located between base and button piece, wherein, when pressing drive button piece moves down, magnet piece can be through second buffer and base and abutment collision;Spring drives button piece to reset and move up, button piece can be through first buffer and upper cover and abutment collision. By the above structure, first buffer and second buffer cooperate to effectively weaken or eliminate the collision sound energy generated by button piece at the end of up-down stroke, so that the collision sound can be greatly reduced or eliminated.

The invention discloses a sensor-based multi-channel bridge measuring device, which comprises a signal acquisition module, a signal conditioning module, a data processing module and a communication interface module, and is characterized in that the signal acquisition module is composed of 8-32 independent bridge circuits, each bridge circuit is connected with a resistance strain type sensor, and a full-bridge or half-bridge structure is adopted; the signal conditioning module comprises a multiplexer, an instrument amplifier and a filter circuit; and the multiplexer adopts a low-on-resistance analog switch. According to the multi-channel bridge measuring device based on the sensor, through multi-channel synchronous measurement (8-32 channels can be selected), the problem that a traditional device is low in single-point measurement efficiency is solved, the detection efficiency is improved, and through multiple anti-interference design (optoelectronic isolation, shielding and a filtering algorithm), the detection efficiency is improved. The problem of industrial field noise interference is solved, and measurement precision is guaranteed.

The application discloses a kind of meridian tire X-ray image disease flaw automatic detection method based on improved YOLO-v5 model, belong to meridian tire detection technical field, including following steps: step 1, meridian tire X-ray image is collected and is segmented processing, uniform resolution, model training sample data are made;Step 2, for the image strip, block missing caused by X-ray machine instability, thereby affecting the detection effect Problem, image restoration processing is carried out;Step 3, design improved YOLO-v5 model, including increasing the fourth detection layer, increase attention module, and improve loss function;Step 4, using meridian tire disease flaw data carries out model training;Step 5, using the model of training completion carries out actual application scene tire disease flaw detection.The application is a kind of automatic detection method, based on improved model can automatically identify multiple disease flaws, and different disease flaws are detected and classified, and detection efficiency and accuracy are higher.

The invention discloses a cross-modal pedestrian re-identification method and device, and relates to the technical field of computer vision. The invention discloses a cross-modal pedestrian re-identification method, and the method comprises the steps: obtaining a to-be-identified visible light image and a to-be-identified infrared image, and carrying out the preprocessing, and obtaining an input image; inputting the input image into the trained pedestrian re-identification network to obtain an identification result; the pedestrian re-identification network comprises a convolutional neural network, a structure perception enhancement module arranged in a shallow layer of the convolutional neural network and a collaborative low-rank decomposition module arranged at the tail end of the convolutional neural network; the structure perception enhancement module is used for extracting shallow layer features; the collaborative low-rank decomposition module is used for extracting deep features. According to the method, the problems of cross-modal feature misalignment and noise interference are effectively solved, and the precision and robustness of cross-modal pedestrian re-identification can be remarkably improved.

The invention relates to the technical field of computer vision, and discloses an image data enhancement preprocessing system for gesture recognition, which comprises an image acquisition unit, a data enhancement unit, a feature extraction unit and a preprocessing output unit. The image acquisition unit receives multi-source input of visible light, infrared and depth sensors and the like, and performs de-weighting and resolution calibration; the data enhancement unit implements geometric transformation and color adjustment based on gesture features; the feature extraction unit identifies contour edges, textures and joint points; and the preprocessing output unit verifies the data quality and packages and outputs the data in a standardized manner. According to the system, through multi-source data fusion, semantic perception enhancement and multi-dimensional feature extraction, the training data quality of a gesture recognition model is remarkably improved, the problems of data isomerism, feature distortion and the like in a traditional method are solved, and the system has the advantages of being high in processing efficiency and adaptability.

This application discloses an on-board oxygen supply device and method, and a vehicle. The on-board oxygen supply device includes an air compressor, which is suspended from the vehicle floor and used for filtering and compressing air; an oxygen chamber molecular sieve assembly, which is connected to the air compressor to receive compressed air and perform gas separation to produce oxygen; and a first oxygen delivery pipe, the inlet end of which is connected to the oxygen outlet of the oxygen chamber molecular sieve assembly, and the outlet end of which is connected to the vehicle's interior cavity. This on-board oxygen supply device suspends the air compressor from the vehicle floor, achieves air separation and oxygen production through the oxygen chamber molecular sieve assembly, and delivers the oxygen to the vehicle's interior cavity through the first oxygen delivery pipe. The structure and layout are reasonable, which can not only stably realize the oxygen production and supply function and ensure the health needs of drivers and passengers, but also effectively reduce equipment operation interference and improve driving comfort.

The invention relates to the technical field of image processing, in particular to an image focusing method and device based on fusion frequency domain information and a storage medium, and the method comprises the steps: obtaining multiple frames of images of a microscope at different Z-axis positions to form an image sequence; extracting a frequency domain definition index and a spatial domain definition index of each frame of image in the image sequence; the frequency domain definition index and the spatial domain definition index are combined, and a multi-modal definition time sequence feature sequence is constructed; inputting the multi-modal definition time sequence feature sequence into a preset improved long-short-term memory network for prediction so as to obtain a definition change trend of a plurality of frames in the future; a frame corresponding to the maximum definition value is located and predicted according to the definition change trend, and the corresponding Z-axis position is determined as the optimal focus point position; and controlling the Z axis of the microscope to move to the optimal focusing point position to complete focusing. According to the scheme of the invention, the robustness and accuracy of focusing are effectively improved.

The invention discloses a numerical control equipment fault prediction system based on edge calculation, and relates to the technical field of equipment state monitoring. Comprising a data acquisition and preprocessing module which is used for acquiring high-frequency and multi-source time sequence data of each sensor of numerical control equipment in real time, performing cleaning by adopting a lightweight preprocessing algorithm, and retaining original time sequence characteristics of the data; and the hierarchical data processing module is used for layering the preprocessed data according to the importance degree of the sensor and the frequency characteristic of the data, and dividing the high-frequency data into core data layers. According to the invention, the data acquisition and preprocessing module adopts a lightweight algorithm to clean data, original time sequence features are reserved, and the problems of noise interference and feature loss in data processing are solved; the hierarchical data processing module is layered according to importance and frequency, high-precision marking of core data layers is achieved, compression of auxiliary data layers is achieved, edge computing power is reasonably distributed, and processing efficiency is improved.

The invention discloses a noise source positioning method and system based on multi-integral-surface acoustic monitoring. The method comprises the following steps: arranging a microphone array on an outermost integral surface, converting a sound pressure signal into a time-frequency domain through short-time Fourier transform, and reconstructing a whole sound field at one time by using a regularization equivalent source method; and then the active sound intensity on the multi-layer nested integral surface is calculated in sequence, and the net sound power is solved, so that the normalized sound power divergence of the adjacent shell layers is obtained. And performing time-frequency accumulation and peak value detection on the divergence spectrum to quickly lock a shell layer where a sound source is located, and determining a three-dimensional coordinate in the shell layer according to an equivalent source energy density extreme value. The device comprises an array module, an acquisition module, a time-frequency conversion module, a reconstruction module, a divergence calculation module and a peak value detection and positioning module, and can complete noise source visualization in real time. The method has the advantages of high spatial resolution, strong noise immunity and millisecond processing speed, and is suitable for online noise source identification and fault diagnosis of vehicles, aerospace and industrial equipment.