14 results about "Daubechies wavelet" patented technology

The invention relates to the technical field of deep learning and load identification, in particular to a deep learning load identification method and system based on bilateral filtering denoising and multi-wavelet feature fusion and a medium, and the method comprises the steps: firstly converting an acquired training data set into an image, and carrying out the preprocessing of bilateral filtering denoising; graying the de-noised image, extracting low-frequency and high-frequency components by using Haar wavelet transform, and extracting low-frequency approximation and high-frequency information in horizontal, vertical and diagonal directions by using Daubechies wavelet transform; pixel unification and normalization are carried out on the feature map, a training set and a test set are divided after category label integers are coded, and a convolutional neural network containing two branches is constructed to extract depth features and splice and fuse the depth features; and finally, extracting fusion features through a full connection layer, and taking sparse classification cross entropy as a loss function to train a CNN model in an off-line manner to obtain a load identification model. The method can improve the accuracy and stability of load identification, and is suitable for various electric equipment load identification scenes.

The invention discloses a multi-modal geologic feature fusion method and a multi-modal geologic feature fusion system based on a discrete wavelet transform (CLIP)-Stable Diffusion model. The method comprises the following steps: reconstructing multi-modal geological data through differential wavelet transformation, extracting a low-frequency trend from earthquake and deposition data by adopting a Daubechies wavelet, capturing high-frequency details from deposition numerical simulation and logging data by adopting a Coiflets wavelet, and vectorizing a geological text through an orthogonal basis matrix; a CLIP-Stable Diffusion fusion model is constructed, a text encoder is utilized to analyze geological semantic features, an image encoder is utilized to extract spatial features, and a U-net diffusion generator realizes cross-modal alignment under the guidance of text conditions through a cross attention mechanism; and adopting a loss threshold termination mechanism constrained by a geological law in diffusion training, and finally outputting fusion data through wavelet inverse transformation. According to the method, cross-scale correlation deficiency and semantic segmentation limitation of a traditional method are broken through, and reservoir modeling precision and exploration efficiency are remarkably improved.

The invention discloses an image small target detection method based on a fusion feature visual model, and relates to the technical field of computer visual detection. And the fusion feature visual detection model performs the following processing on an input image and outputs a detection result: adopting adaptive Daubechies wavelet transform to enhance high-frequency details, combining multi-scale feature fusion with a double attention mechanism to enhance small target feature expression, utilizing regression calculation to predict bounding box coordinates, and realizing category probability distribution prediction based on a Softmax function. A multi-task loss function and an Adam optimizer are adopted for end-to-end training, and the model performance is improved by balancing classification and regression loss. The method effectively solves the problems that small target features are deficient and are susceptible to background interference, significantly improves the detection precision while maintaining the calculation efficiency, and is especially suitable for unmanned aerial vehicle patrol and other practical application scenes.

The invention discloses a perception method based on multi-semantic space attention and Daubechies wavelet double-branch Mama, and belongs to the field of hyperspectral image processing. The problem that in the existing Mama-based model feature extraction process, the capacity of capturing multi-scale local structures and direction sensing information is insufficient is solved. The method comprises the following steps: inputting a hyperspectral image; projecting the spectral vector to an embedding space through an embedding layer to obtain an embedding feature; inputting the embedded features into an encoder, wherein the encoder comprises an SMSAMama branch, a DWTMama branch and a self-adaptive feature fusion module; the SMSAMama branch is used for extracting spatial features; the DWTAMba branch is used for extracting spectral features; the adaptive feature fusion module performs weighted integration on the spatial features and the spectral features by using randomly initialized fusion weights; and inputting the integrated features into a segmentation head to generate a final perception result. The method is used in agricultural monitoring and urban planning fields.

The invention provides a top tension riser damage identification and positioning method based on wavelet packet transformation, and belongs to the technical field of ocean structure intelligent monitoring. Firstly, a vibration acceleration response signal of a preset node of the riser is collected and preprocessed; using a 6-order multi-Bezi wavelet as a mother wavelet, setting a four-layer wavelet packet decomposition layer number, and iteratively decomposing the preprocessed signal through a Mallat algorithm to obtain 16 frequency band component signals; secondly, calculating wavelet packet component energy of each node under 16 frequency bands by utilizing the sensitivity of the method to structural rigidity damage based on an arch bridge damage identification method based on the energy of the Singxiang wavelet packet; solving the wavelet packet energy curvature of each node through a variable quadratic difference value; and finally, the difference between the curvature and the curvature of the same node of the healthy riser is obtained to obtain an IWPECD index, the distribution rule of the index along the length of the riser is analyzed through five-point moving average smoothing, the node corresponding to the peak value is the damage position of the riser, and effective damage positioning of the jacking tension riser is achieved.

The invention discloses an integrated learning-based unmanned aerial vehicle electrical appliance state evaluation and fault prediction method and device, and a medium. The method comprises the steps of collecting operation data; daubechies wavelet denoising is adopted, and transient fault features are reserved; searching an optimal LightGBM hyper-parameter in a hyper-parameter space by utilizing Bayesian optimization, training the model, and outputting an operation parameter predicted value at the next moment; and calculating the residual error of the predicted value and the real value, and judging abnormity and positioning a fault subsystem if the residual error exceeds the limit. The de-noising link is based on the thinnest-layer detail coefficient self-adaptive threshold shrinkage, and the Bayesian link adopts a Gaussian process-TPE acquisition function. According to the invention, the fault positioning accuracy can be effectively improved. The whole link does not need artificial features, online scheduling is supported to participate in embedded deployment, and the false alarm and operation and maintenance cost is remarkably reduced.

The invention relates to an indwelling catheter blockage early warning method and system based on pressure waveform analysis, and relates to the technical field of monitoring early warning, and the method comprises the following steps: collecting a multi-source monitoring data set comprising original pressure waveform data, instantaneous flow velocity data and electrocardio synchronous signal data; performing four-layer multi-scale decomposition processing based on a Daubechies wavelet basis function on the original pressure waveform data to generate a multi-scale waveform component set; constructing a catheter downstream vascular bed fluid mechanical impedance model by combining the basic information of the patient, and deducing and generating a catheter downstream vascular bed impedance change index; dynamic baseline drift modeling is carried out through a sliding time window, and a self-adaptive catheter blockage risk reference threshold value is set; constructing a patient specific risk factor graph structure, performing time sequence prediction analysis on the feature sequence through a recurrent neural network, and generating a future risk evolution trend index; and a third-level early warning signal is generated. The method has the beneficial effect that the sensitivity and the specificity of early warning of catheter blockage are remarkably improved.

The invention relates to the technical field of environment and climate science, and discloses a seasonal infectious disease diffusion risk prediction system based on big data, and the system comprises the steps: collecting multi-source heterogeneous data, and fusing the data into a multi-dimensional feature library; performing multi-scale wavelet transform decomposition by adopting a Daubechies wavelet basis function to form a layered feature representation library; carrying out fractal geometric constraint scale self-similarity mapping on the low-frequency seasonal components by adopting a Mandelbrot set variant, and generating a mapped seasonal vector; embedding infectious disease dynamic constraints, and optimizing parameters through a projection gradient method and a mixed loss function; multi-quantile uncertainty distribution is generated through Monte Carlo sampling, and multi-scale risk assessment is carried out; and according to the harmonic oscillation phase and fractal feedback, dynamically adjusting an early warning threshold and a resource scheduling strategy, and outputting a risk early warning report. According to the invention, cross-seasonal risk prediction is realized through a multi-scale wavelet harmonic coupling framework.

The present application relates to a kind of based on pressure waveform analysis's indwelling catheter blockage early warning method and system, it is related to monitoring early warning technical field, including acquisition contains original pressure waveform data, instantaneous flow rate data and electrocardiogram synchronous signal data multisource monitoring data set;Original pressure waveform data is executed based on Daubechies wavelet base function four-layer multiscale decomposition processing, generates multiscale waveform component set;Combining patient basic information constructs catheter downstream vascular bed fluid mechanics impedance model, deduces and generates catheter downstream vascular bed impedance change index;Through sliding time window, dynamic baseline drift modeling is carried out, and adaptive catheter blockage risk benchmark threshold is set;Construct patient-specific risk factor graph structure, through recurrent neural network, time series prediction analysis is carried out to feature sequence, generates future risk evolution trend index;Generate three-level early warning signal.The present application has beneficial effect for significantly improving the sensitivity and specificity of catheter blockage early warning.

The application discloses a three-phase low-voltage line single-phase grounding fault monitoring and protection method and system in the field of power system relay protection and electrical monitoring, and the application takes the bus zero sequence current amplitude over-limit as the starting criterion of the system, obtains the transient current and voltage at the bus and at the outlet of each line, selects the fault line by the Daubechies wavelet modulus extreme value comparison method of the transient current traveling wave, further calculates the Hermitian wavelet energy of each phase current of the fault line to select the fault phase, triggers the bypass grounding switch to act by the microprocessor, and makes the fault current discharge to the ground through the current-limiting resistor without affecting the operation of the power grid line, thereby protecting the operation stability of the system. The protection method provided by the application has the advantages of fast detection speed, low cost, simple device, higher reliability and better safety compared with the existing single-phase grounding protection device.

The present application relates to the field of communication signal processing, in particular to a kind of air defense disaster early warning alarm real-time control system based on Beidou third-generation communication, system uses Daubechies wavelet family to carry out 5 layers discrete wavelet decomposition, and total satellite signal delay is decomposed into different scale coefficients. By accurately controlling the boundary scale, the system can effectively separate ionospheric error and tropospheric error. By inserting zero values between filter coefficients, information loss caused by downsampling operations is avoided. The system also adaptively selects the optimal wavelet basis function according to different atmospheric error characteristics. It maintains stable correction effect under extreme weather conditions and builds an efficient and practical air defense disaster real-time early warning control method.

A kind of finger vein and finger knuckle print dual-mode identification method based on canonical correlation analysis, comprising: Daubechies wavelet pre-processing is carried out to the finger vein image and finger knuckle print image for training, to obtain the low-frequency image of finger vein image and finger knuckle print image respectively, K-L transform is used to extract features from two groups of low-frequency images respectively, then correlation criterion function between two groups of feature matrices is established, two sets of canonical projection vectors are obtained according to criterion function, finally the canonical correlation characteristics are fused by given feature fusion strategy and applied in classification identification.The beneficial effects of the present application are: the present application obtains the feature information of two modes by dimension reduction, reduces the recognition calculation amount, improves the recognition accuracy, introduces more class information into the criterion function more fully, not only reaches the purpose of information fusion, but also effectively eliminates the redundant information between features, and makes the classification performance greatly improved.

The invention belongs to the field of numerical control machining and manufacturing of complex thin-walled parts, and provides a thin-walled part milling state evaluation method and system based on sliding window-wavelet transform. The method comprises the following steps: collecting a milling force and an acceleration signal, denoising and normalizing, segmenting by adopting a sliding window, carrying out wavelet transform on each data block, carrying out multilayer decomposition by selecting a Daubechies wavelet basis function, identifying a sensitive frequency band related to flutter and tool wear based on an energy threshold, and reconstructing and generating a two-dimensional time-frequency spectrogram through a soft threshold; a convolutional neural network is utilized to learn features and establish a recognition model, and real-time evaluation and early warning of steady-state machining, flutter, cutter failure, cutting overload and other states are realized. According to the method, a flutter mechanism and time-frequency analysis are fused, the recognition sensitivity and accuracy of non-stable and transient states are remarkably improved, the method is suitable for intelligent manufacturing of complex thin-wall parts such as aero-engines, and effective support is provided for machining quality control.

This invention relates to the fields of building engineering technology and intelligent monitoring technology, specifically disclosing a BIM-based building construction data management system. The system includes a data acquisition module, a soil stability analysis module, a settlement trend identification module, a foundation condition judgment module, and a risk linkage control module. The system deploys high-precision soil negative pressure sensors and micro-settlement monitoring instruments to collect foundation-related data in real time. It then uses methods such as sliding window mean processing, Daubechies wavelet transform, and empirical mode decomposition to extract soil negative pressure change characteristic values ​​and cumulative settlement trend characteristic values. Furthermore, it integrates these two types of features into a comprehensive feature vector, inputs it into a support vector regression model for foundation risk level assessment, and combines it with the BIM platform to achieve construction progress linkage and hierarchical early warning control.