281 results about "Dynamic modulation" patented technology

The invention relates to the technical field of fault diagnosis, in particular to an unmanned aerial vehicle motor fault diagnosis method based on multi-modal data fusion, and the method comprises the steps: employing a parallel five-layer one-dimensional CNN to extract deep modal features of vibration and current signals, employing a 64 * 16 convolution kernel for a first layer, employing a 3 * 1 convolution kernel for a subsequent layer, and carrying out the pooling; the features are integrated layer by layer through a multi-layer feature fusion module, a weight map is generated through 3 * 3 convolution and Sigmoid, and the features are dynamically modulated and then spliced and fused; the fusion feature dimension is adjusted to adapt to Transform, position coding is introduced, and global dependence is modeled by using a multi-head self-attention mechanism; and finally, nine types of faults are classified through a deep neural network full connection layer. The invention aims to solve the problems that traditional single-modal fault diagnosis feature acquisition is not comprehensive, modal difference is neglected in early-stage fusion in multi-modal fusion, inter-modal associated information mining is lacked in late-stage fusion, and CNN long-distance dependence modeling capability is insufficient.

The invention belongs to the technical field of cleaning and dirt prevention and removal, and discloses a control system and a control method of an ultrasonic cleaning device, stable energy coupling and dead-corner-free cleaning are realized through double-closed-loop driving and dynamic modulation in an ultrasonic cleaning link, and the uniformity of tank liquid cavitation intensity is remarkably improved. Meanwhile, control strategies are customized in the aspects of a rinsing link, a water cutting and drying link, liquid level control, manipulator control and the like, so that the defects of high energy consumption, incomplete drying, low rinsing efficiency, manipulator conflict and the like in the prior art are overcome while the surface and the interior of the workpiece are thoroughly cleaned; and the cleaning effects of high cleanliness, complete drying and no physical or chemical damage are ensured.

The invention discloses a transparent element tiny defect double-stage detection method and system. A double-stage defect detection network framework generated based on pseudo labels is designed. In the first stage, pseudo-label generation of transparent element data is carried out, an unsupervised detection method and a pseudo-label extraction strategy are adopted, and a spatial pyramid balance feature fusion-adaptive dynamic modulation convolutional network is provided in the aspect of the unsupervised detection method. The network is combined with pyramid balance feature fusion and adaptive dynamic modulation convolution, so that the feature extraction capability of an unsupervised detection network on a multi-scale target and the effectiveness and flexibility of coping with more complex data features are enhanced. In the aspect of a pseudo label extraction strategy, a pseudo label dynamic mining strategy based on fusion of threshold segmentation and RPN proposal is provided, and pseudo labels of various defects can be directly generated from an unsupervised network detection result. And the second stage is defect detection based on a pseudo label. According to the network, the performance of the supervised defect detection network is remarkably improved.

The invention discloses a power grid broadband oscillation tracing method and system based on a modal dynamic modulation graph network, and the method comprises the steps: inputting an oscillation data sequence of a power grid and a power system graph into the modal dynamic modulation graph network to obtain an oscillation source positioning result based on the power system graph; the processing of each stage of modal dynamic modulation graph module in the modal dynamic modulation graph network comprises the following steps of: modulating an oscillation data sequence through a modal modulator and then performing characteristic modal decomposition; key features are extracted through a convolution layer and a pooling layer; dynamic modulation is carried out based on a Prony dynamic modulator, and feature blocks are segmented; key features are extracted through a convolution layer and a pooling layer; extracting node features through a graph network; and realizing power grid broadband oscillation traceability by using a classifier according to node features. The method aims at reducing interference of other information on broadband oscillation information, data are dynamically analyzed from the two dimensions of time and space, and broadband oscillation source positioning is rapidly and accurately achieved.

The invention provides a geometric parameter measurement method based on underwater structure crack pixel-level semantic segmentation, and the method comprises the steps: constructing a multi-level feature extraction network based on ResNet-34, providing a structured feature pyramid for the subsequent context modeling and interference suppression, modeling illumination distribution features through a multi-head self-attention mechanism, and carrying out the modeling of the illumination distribution features through a multi-head self-attention mechanism. Regional scattering noise is extracted in combination with a cavity convolution pyramid, an environment context tensor is generated through fusion, a learnable gating mechanism is introduced for a decoding stage, dynamic modulation of jump connection characteristics and interference area response suppression are achieved, in addition, an edge perception auxiliary loss function is adopted, topological continuity of crack edges is enhanced, and finally, the method is applied to the field of jump connection. High-precision pixel-level segmentation of the underwater crack area can be realized without prior illumination correction, and the detection robustness and segmentation integrity in a complex environment are improved.

The invention discloses a visible light communication data transmission method based on dynamic modulation, and particularly relates to the field of visible light technology communication. The method comprises the following steps: performing nonlinear dynamic feature extraction on an actual light intensity signal emitted by a light emitting diode to generate a spectral response feature set; performing segmented analysis on the frequency domain distortion evolution trend based on the frequency spectrum response feature set to obtain a plurality of frequency spectrum distortion subintervals; performing local modulation parameter optimization on each spectrum distortion subinterval to obtain a dynamic local modulation parameter set; mapping the dynamic local modulation parameter set into a modulation driving strategy sequence, and generating a dynamic modulation waveform signal sequence; performing detection and time-frequency feature extraction on the dynamic modulation waveform signal sequence to obtain a target signal sequence after noise suppression; and the target signal sequence is demodulated, complete data is recovered, visible light communication data transmission is completed, and the optical communication signal quality and the data recovery capability in a complex modulation environment are effectively improved.

The invention relates to the technical field of data processing and mode recognition, in particular to a bearing fault diagnosis method based on multi-scale feature fusion, which comprises the following steps: fusing multi-source data such as vibration, acoustic emission and rotating speed, performing angle domain resampling by using rotating speed data, generating a two-dimensional order spectrogram, and stacking to construct a three-dimensional working condition information tensor; a master-slave modulation heterogeneous neural network is adopted, high-dimensional spatial-temporal features are extracted through a main branch three-dimensional convolutional network, time sequence details are extracted from an original sequence through an auxiliary branch one-dimensional convolutional network, affine transformation parameters are generated, and dynamic modulation is achieved on the high-dimensional features; and the output state vector is mapped to a fault evolution knowledge graph, probability prediction is carried out through a graph attention network and by introducing a Monte Carlo discarding mechanism, a probability mean value is calculated as a fault classification result, and the diagnosis confidence is quantified by a probability variance. According to the invention, through multi-scale feature fusion and dynamic modulation, the problem of insufficient feature discrimination caused by scale mismatch under variable working conditions is solved.

The invention discloses an intelligent data mining system and method based on big data service, and aims to effectively fuse instant interests and historical preferences of users by introducing a short-term intention feature vector and a long-term user preference portrait coding vector and utilizing a fine-grained semantic comparison network to realize deep interaction between the short-term intention feature vector and the long-term user preference portrait coding vector. On the basis, the recall quantity proportion of the first recall channel and the second recall channel is designed to be dynamically modulated according to the comparison result between the short-term intention feature vector and the long-term preference portrait coding vector. In this way, the recall weights of different types of commodities or contents can be adaptively adjusted according to the consistency or difference of the current demands and historical interests of the users in different scenes. The dynamic modulation mechanism significantly improves the adaptability of the recommendation system to diversification and real-time performance in a complex scene, and is an important technical breakthrough for future intelligent services.

The invention provides an optical cross system based on silicon-based liquid crystal and a control method, and relates to the field of optical elements, and the optical cross system comprises an optical fiber array, a micro lens array, a collimating lens, a polarization conversion module, a dispersion element, a focus lens group and a liquid crystal spatial light modulator. Optical signals are coupled through the micro-lens array, a divergence angle is compressed, and parallel light beams are formed through the double-agglutination collimating lens; the polarization conversion module separates the light beam into P / S polarized light and unifies the polarization state; the dispersion element separates light beams according to wavelength space, and the focus lens group forms an elliptical Gaussian light spot on the liquid crystal spatial light modulator; the liquid crystal spatial light modulator divides a light spot into a plurality of independent control areas in the long-axis direction, and single-wavelength multi-port output, power proportion regulation and control and monitoring signal extraction are achieved by dynamically modulating the reflection phase of each area. The control method comprises the steps of optical path processing, regionalized phase modulation and real-time monitoring feedback. According to the invention, flexible routing, power balance and lossless monitoring of wavelength channels can be realized.

The invention relates to the technical field of beacon light control, in particular to a beacon light anti-interference control method and system based on environmental perception and dynamic modulation. The method comprises the following steps: extracting an ambient light reference based on adaptive weight sliding filtering; in combination with target detection and trajectory prediction, interference pre-judgment features are obtained; on the basis of an evidence theory, analyzing comprehensive interference evaluation characteristics through the ambient light reference, the interference pre-judgment characteristics and the electromagnetic interference characteristics; fusing meteorological environment perception and influence features and self health state features, and evaluating equipment operation risks; and based on modulation signal optimization characteristics, the comprehensive interference evaluation characteristics and the equipment operation risk, fuzzy control and rule reasoning are utilized to realize dynamic output of anti-interference control parameters. In order to solve the problem that the signal recognition rate is reduced due to the fact that a beacon light is easily affected by multiple factors in a complex ocean navigation channel environment, data are obtained through multi-dimensional environment perception, and dynamic modulation control is achieved after feature extraction and fusion.

The invention relates to the technical field of optical fiber sensing, in particular to an optical fiber temperature measurement method and system, and the method comprises the steps: injecting a multi-scale dynamic modulation coding laser signal into a to-be-measured optical fiber and a reference optical fiber respectively, collecting two paths of optical signals, and generating frequency domain data containing modulation coding characteristics through fast Fourier transform; constructing a virtual reference channel, fusing pre-stored historical data through digital matched filtering and self-organizing mapping, and realizing stable output of a reference signal; adaptive calibration is performed on data to be measured by using extended Kalman filtering in combination with fuzzy logic, and calibrated phase data accurately reflecting temperature change is output; and finally, outputting high-precision temperature distribution data based on the temperature inversion model and multi-dimensional correlation analysis. The temperature measurement precision and the system robustness are remarkably improved, and the method is suitable for the fields of industrial and environmental monitoring and the like.

The invention discloses an ultra-high-speed imaging device based on acousto-optic filtering modulation. The ultra-high-speed imaging device comprises a spectrum modulation system, a spectrum imaging system and a data processing system. The spectrum modulation system utilizes an acousto-optic tunable filter (AOTF) to dynamically modulate broadband continuous spectrum light, narrow-band illumination light which rapidly changes along with time is generated, and time information of a dynamic scene is effectively mapped to a spectrum dimension. And the spectral imaging system images the spectral coded dynamic scene to the hyperspectral camera to realize acquisition of spectral space-time information. The data processing system performs spectral crosstalk correction, channel separation and demosaicing reconstruction on the acquired original mosaic image, and recovers a time sequence image according to a spectrum-time mapping relation, thereby realizing single-shot superspeed imaging in a dynamic process. The device has the advantages of being simple and compact in structure, flexible in time window, high in imaging fidelity and the like, and has wide application prospects in the fields of rapid dynamics such as microfluidics, laser processing and ultrafast physics.

The invention discloses a dynamic digital twinning three-dimensional Gaussian splash rendering system and a dynamic digital twinning three-dimensional Gaussian splash rendering method. According to the system, a static scene is reconstructed by utilizing a three-dimensional Gaussian splashing technology through a semantic scene reconstruction module, semantic information is associated to Gaussian primitives in combination with spatial alignment and semantic injection, and a model with a semantic identifier is generated; multi-source dynamic data are fused and converted into a dynamic data field which can be accessed by a GPU through a multi-mode spatio-temporal data processing module; and through a rendering attribute dynamic modulation module, screening a target primitive according to the semantic identifier, and directly modulating internal rendering attributes such as color, transparency or shape of the target primitive based on a data field query value, thereby realizing internal rendering and visualization of dynamic data. According to the method, the problems that in the prior art, rendering reality and real-time performance are difficult to give consideration to and data and scene fusion is superficial are solved, and dynamic digital twinning rendering with high reality and strong immersion is realized.

The invention discloses a driver state remote monitoring collaborative alarm platform and method based on the Internet of Things, and relates to the technical field of the Internet of Things, and the method comprises the steps: carrying out the channel information entropy analysis of collected state data, adaptively cutting off a low-information-amount channel, carrying out the nonlinear feature mapping to obtain a mapping feature sequence, constructing a space-time coupling information tensor, and carrying out the self-adaptive cutting of a low-information-amount channel; the method comprises the following steps: calculating a multi-mode cooperative surge factor, executing tensor decomposition to obtain a decomposition factor, reconstructing an estimation tensor and calculating a reconstruction error, calculating a multi-mode entropy surge factor, generating an abnormal score, counting and generating a dynamic threshold according to the abnormal score in a sliding window, and judging a real-time score to generate an alarm trigger signal. Through adaptive dimensionality reduction, hybrid mapping, dynamic modulation CP decomposition and adaptive reconstruction error evaluation based on channel information entropy, real-time efficient processing and deep multi-modal fusion of multi-source data are realized, noise and redundancy are effectively eliminated, abnormal features are amplified, space-time coordination is accurately described, and dynamic adaptive anomaly judgment is realized.

The invention relates to the field of metal plastic forming control, and discloses a self-adaptive control system and method for forging and pressing aviation precision forgings, and the system comprises an online dynamic model identification module which identifies a process fingerprint representing the characteristics of a current blank at the initial stage of machining; the target curve self-adaptive regulator is used for regulating a reference energy dissipation rate curve according to the fingerprint and generating a personalized reference curve; the model prediction controller performs rolling optimization by using continuously updated model parameters and taking the reference curve as a tracking target, and outputs an optimal speed instruction; and the event triggering and constraint dynamic modulation module is used for actively tightening the operation constraint applied to the controller before a forging pressure sudden increase event is pre-judged through a parallel monitoring model prediction error. According to the method, through combination of target self-adaption and constraint dynamic modulation, control over the forging and pressing process is achieved, forging pressure overshoot can be effectively restrained, and the forming precision and batch-to-batch quality consistency of forge pieces are remarkably improved.

The invention provides a double-optical-comb-cavity modulated gas measurement system and method based on electro-optical modulation, in the use process, a detection optical comb and a reference optical comb which are coherent are generated through a double-optical-comb output module, the detection optical comb enters a resonant cavity to interact with gas to be measured, and the reference optical comb directly reaches a data acquisition and analysis module; the data acquisition and analysis module is communicated with the resonant cavity and the double-optical-comb output module, and is used for receiving the reference optical comb and the detection optical comb output from the resonant cavity, and performing component and concentration analysis on the to-be-detected gas according to the detection optical comb and the reference optical comb; in the measuring process, the electro-optical modulation module is used for acquiring the detection optical comb output from the resonant cavity in real time, and the optical field of the detection optical comb in the resonant cavity is dynamically modulated based on the detection optical comb, so that the optical comb mode is matched with the free spectral range of the resonant cavity in real time, and efficient coupling is formed in the resonant cavity; the influence of the gas dispersion effect on the measurement precision is eliminated.

The invention provides a video content enhancement method for a low-illumination environment, and the method comprises the steps: achieving the data preprocessing based on an original low-illumination video frame sequence through frame synchronization, color space conversion and local brightness analysis, generating a noise sensitivity thermodynamic diagram through multi-feature unsupervised learning, and constructing a noise perception gating mechanism through the combination of affine transformation. Dynamic modulation of the characteristic channel is realized; in the multi-scale network structure, a channel attention module is used for carrying out layer-by-layer self-adaptive adjustment on a noise sensitive area; a basic illumination image and an edge enhancement image are generated through double-branch decoding, and then weighted fusion is carried out in combination with a noise thermodynamic diagram, so that brightness balance and detail enhancement are realized; a noise smoothing regular term is introduced during end-to-end training, so that the network achieves dynamic balance between an enhancement effect and noise control.

The invention discloses a display light color detection system, and relates to the technical field of low-brightness display light color detection, and the system comprises a dynamic modulation module which is configured to carry out the high-frequency pulse modulation of a dark-state signal of a detection object, and is used for enabling the dark-state signal to carry a modulation component of a characteristic frequency; the time sequence division acquisition unit comprises a high-speed optical shutter assembly and a detector, and the high-speed optical shutter assembly synchronously switches the on-off state of an optical path according to the characteristic frequency; according to the invention, through high-frequency pulse modulation of dark state signals, time-sequence-divided acquisition of noise and signals and time-frequency domain combined decoupling, traditional hardware limitation is broken through, ultra-low brightness signals and stray light interference are effectively separated, 0.01 cd / m < 2 >-level high-precision detection is realized, the detection efficiency is improved by more than 7 times, and the misjudgment rate is significantly reduced. The problems of missing detection of a novel display technology and the yield of an automatic production line are solved, and an active coding-time sequence calibration-cross-domain cooperation framework has remarkable innovativeness and industrial value.

The invention relates to the technical field of frequency converters, and discloses a circuit for suppressing current oscillation of a frequency converter, which comprises a direct current bus voltage oscillation separation unit comprising an arithmetic unit and a PWM (Pulse Width Modulation) modulator; the PWM modulator is configured to calculate the DC bus voltage by using the arithmetic unit, and modulate the calculated DC bus voltage to obtain a modulation signal and a dynamic modulation ratio; wherein the dynamic modulation ratio is a factor which reversely changes along with the fluctuation of the direct-current bus voltage; a comparator configured to compare the corrected given voltage signal and the modulation signal to output a control signal; the corrected given voltage signal is obtained by correcting the given voltage signal according to the dynamic modulation ratio. In the circuit, the fluctuation voltage affects the given voltage, so that the corrected given voltage counteracts harmonic waves of the real-time direct-current bus voltage, and current oscillation is eliminated. The invention further discloses a method and device for suppressing current oscillation of the frequency converter, the frequency converter and a computer readable storage medium.

The invention relates to the technical field of lighting systems, and provides a remote LED lighting method and system based on the Internet of Things. Obtaining an initial parameter of a lighting system through the Internet of Things, and generating a multi-dimensional lighting function based on the initial LED parameters of the lighting node and the neighbor node corresponding to the LED equipment; inputting the LED parameters of the illumination node and the neighbor node thereof at the current moment into an illumination function, and determining the state variable of the illumination node at the next moment; dynamically modulating a state variable according to the power supply parameter and the LED parameter, and determining the signal frequency of a current signal passing through the LED equipment; and controlling the LED equipment to illuminate based on the state variable and the signal frequency. By determining the multi-dimensional illumination function, dynamically predicting the state variable and modulating the current signal frequency, the adaptability and flexibility of illumination are improved, the luminous efficiency of the illumination system is enhanced, and the energy consumption is reduced.

The invention discloses a geometric error identification and compensation system and method for a numerical control machine tool, and belongs to the technical field of numerical control machine tools, and the method comprises the steps: collecting the position, temperature and cutting force data of the machine tool in real time, and calculating the components of a position error, a thermally induced equivalent error and a force induced equivalent error to construct a multi-dimensional state vector; according to the method, a compensation value is calculated through a dynamic compensation model, the core of the method is that a dynamic modulation factor is introduced, the factor integrates the heat effect, the force effect and the multi-axis coupling effect to dynamically adjust the compensation weight, in addition, a grading trigger compensation strategy based on comprehensive error amplitude evaluation is designed, and the compensation accuracy is improved. The reconstruction compensation can update model parameters on line to cope with long-term changes of the system, intelligent distribution of compensation resources and self-adaptive adjustment of system changes are achieved, time-varying errors are effectively restrained, and precision, efficiency and robustness are taken into consideration.

The invention discloses a dynamic encryption transmission method and system for sound-light alarm signals, and relates to the field of data encryption, and the method comprises the steps: obtaining a dynamic environment entropy value; generating a pseudo-random chaotic sequence; obtaining a helium-oxygen mixed gas density parameter in the dynamic environment entropy value; segmenting a preset standard alarm waveform into a plurality of signal waveforms in a time domain, and generating a distorted signal waveform according to the helium-oxygen mixed gas density parameter; performing encryption modulation on the distorted signal waveform to generate an encrypted distorted signal waveform; determining a sending sequence and a sending interval of the encrypted distortion signal waveform by combining a pseudo-random chaos sequence based on an impedance fluctuation value and background acoustic spectrum characteristics in the dynamic environment entropy; generating a dynamic modulation signal based on the transmission sequence and the transmission interval; the dynamic modulation signal is sent to a water surface console through an umbilical cable; and receiving the guide pulse, performing phase conjugation processing on the guide pulse, generating an echo verification signal, and transmitting the echo verification signal back to the water surface console.

The invention discloses a six-dimensional force sensor dynamic decoupling and coupling processing method, and relates to the technical field of mechanical measurement sensing detection, and the method comprises the steps: building a dynamic decoupling relation of six-dimensional mechanical component and elastomer strain response coupling mapping based on an original voltage signal and a static calibration parameter of a six-dimensional force sensor; dynamically modulating the mechanical compensation factor of the dynamic decoupling relation through the environmental influence factor; according to the modulated mechanical compensation factor, performing real-time suppression and compensation on coupling deviation caused by environmental disturbance, and outputting a six-dimensional mechanical component subjected to environmental compensation; performing high-precision decoupling processing on the six-dimensional mechanical component subjected to environment compensation to generate a corrected six-dimensional mechanical component; performing closed-loop online updating on the mechanical compensation factor of the dynamic decoupling relation by using the corrected six-dimensional mechanical component; according to the invention, the coupling deviation caused by environmental disturbance is suppressed and compensated in real time, and the defects of the six-dimensional force sensor in the aspects of decoupling precision and measurement stability are improved.

The invention discloses a container damage detection method based on improved UPAD-YOLO, and relates to the technical field of container damage detection, and the method comprises the following steps: image collection: obtaining an image data set containing three types of defects of recess, broken hole and structure damage; preprocessing data, dividing a data set and labeling defect information; constructing a model, and embedding a dynamic modulation convolution module into the backbone network; feature enhancement: introducing a parallelization block perception attention module; feature fusion: integrating a self-adaptive spatial feature fusion module on a detection head; according to the method, multi-scale defect features can be effectively extracted, the detection precision and the anti-interference capability are improved, the problems that the multi-scale defect detection precision is limited and the detection effect is poor under a complex background are solved, and the method is suitable for container damage real-time detection in a large-scale logistics scene.

The invention discloses a vision-language-action robot control method and system based on task stage semantic enhancement, and the method comprises the steps: firstly constructing a task stage semantic system, and disassembling a task into stages with definite semantics; constructing a task stage observation module based on a pre-trained vision-language model, deducing a task state and abnormality in real time, and generating stage semantic features; modulation parameters are generated through a lightweight network, and dynamic modulation is carried out on the vision-language features to strengthen the vision feedback weight; and finally, fusing the modulated characteristics and the body state characteristics, and outputting a control instruction through an action generation network to form closed-loop control. According to the method, the multi-modal information weight can be dynamically balanced, task abnormity can be identified in real time, error correction is triggered, the false completion phenomenon is remarkably reduced, the task execution reliability, robustness and generalization ability of the robot in a complex dynamic environment are improved, and the method is suitable for various robot systems needing high-credibility operation.

The invention discloses an electric hot melting sleeve connection monitoring method for a polyurethane directly-buried thermal insulation pipe, and belongs to the technical field of electric hot melting sleeve installation monitoring. The method specifically comprises the steps that distributed electrode sets are embedded into annular contact surfaces on the two sides of an electric hot melting sleeve, and an electric coupling network is formed by the distributed electrode sets and a conductive enhancing structure in a polyurethane thermal insulation layer; during electrifying and heating, applying a broadband alternating current signal which is dynamically modulated and covers a dielectric relaxation frequency band of a melting interface to the electrode group; collecting electrode pair voltage response signals in real time, and resolving a complex impedance spectrum containing real part and imaginary part impedance components; analyzing real part impedance non-linear attenuation characteristics, and if the real part impedance non-linear attenuation characteristics do not accord with a rule, analyzing an imaginary part impedance phase angle fluctuation rule, extracting dielectric anomaly features, matching the dielectric anomaly features with a preset model, and performing early warning if the dielectric anomaly features exceed a tolerance range; and finally, positioning the spatial distribution of the fusion defects based on the impedance response of the electrode group.

The present application discloses and describes neurostimulation systems and methods that include, among other features, (i) neural stimulation through audio with dynamic modulation characteristics, (ii) audio content serving and creation based on modulation characteristics, (iii) extending audio tracks while avoiding audio discontinuities, and (iv) non-auditory neurostimulation and methods, including non-auditory neurostimulation for anesthesia recovery.

The invention relates to the technical field of emotion recognition, and discloses an emotion recognition method and system based on visual and auditory synergy, and the method comprises the steps: obtaining an original video stream and an original audio stream, introducing an explicit uncertainty quantification mechanism for a visual mode and an auditory mode, extracting features, and meanwhile, carrying out the recognition of the visual mode and the auditory mode. An uncertainty sequence that assesses its reliability is generated in parallel. Furthermore, the uncertainty sequence is used as a core guiding signal, and a cross-modal cross attention fusion process is dynamically modulated. The mechanism enables the model to adaptively suppress the contribution of an unreliable mode, and focuses on the interaction of high-quality signals, thereby effectively suppressing noise propagation, realizing robust multi-mode information complementation, and finally improving the accuracy of emotion recognition.

The invention discloses a wide-spectrum electromagnetic wave modulation method based on silver-based chalcogenide, and belongs to the technical field of electromagnetic wave modulation. According to the invention, the abrupt-change modulation of 1GHz-300THz wide-spectrum electromagnetic waves is realized based on the electronic phase change characteristic of the metal insulator under the triggering of the silver-based chalcogenide under the characteristic external field, and the resonant structure design is combined, so that the functions and applications of electromagnetic wave switching, electromagnetic wave intelligent shielding, dynamic information communication encryption, multi-spectrum camouflage and passive electromagnetic wave protection are realized. The main conception is that a silver-based chalcogenide material with an electronic phase change characteristic is introduced as a sensitive medium, and mutation of electromagnetic wave optical characteristics covering microwave, terahertz and infrared frequency bands caused by electronic phase change of the silver-based chalcogenide is triggered by utilizing characteristic temperature, characteristic pressure and characteristic current. According to the invention, the technical barrier of wide spectrum response and multi-mode dynamic modulation is broken through, and the modulation effect can be cooperatively regulated and controlled by the components of the silver-based chalcogenide material and the structure of the resonant array.

The invention discloses a newborn health monitoring system based on multi-modal data fusion, and belongs to the technical field of medical monitoring. According to the system, aiming at the problem of misjudgment caused by complex background interference and newborn development difference, a background perception feature decoupling enhancement network is constructed, a dual-channel architecture is matched with an adversarial loss function, foreground skin and background features are forcibly separated, and weak pathological features are enhanced in combination with a color space attention mechanism. Meanwhile, a development stage self-adaptive classification network is utilized, the gestational age serves as prior information to generate development codes, and dynamic modulation and weighted evaluation are conducted on fusion features through a premature infant and full-term infant double-branch classifier. According to the invention, the influence of environmental noise is effectively eliminated, and differentiated accurate health state recognition and risk early warning are realized.