594 results about "Compressed sensing" patented technology

The invention discloses a precise interference avoidance method and device in a radio system, and relates to the field of signal processing, and the method comprises the steps: constructing a sensing matrix through sensing node data, and capturing a transient interference signal through aperiodic scanning; performing tensor decomposition on the signal data to extract time domain, frequency domain, space domain and modulation domain features, constructing a dual-mode spectrum analysis model, reconstructing an instantaneous spectrogram by using compressed sensing, and predicting an interference mode through LSTM; after the instantaneous spectrogram and the predicted interference graph are fused, threat assessment is carried out through a multi-stage interference classification model; according to the interference category and the threat level, beam forming is optimized, adaptive null is generated, a power density optimization model is constructed, and the transmitting power is dynamically adjusted; an anti-interference frequency hopping sequence is generated based on a chaotic mapping algorithm, and spectrum camouflage and tracking interference resistance are realized. The method has the advantages that accurate identification and dynamic avoidance of interference are realized through multi-dimensional perception, intelligent prediction and adaptive beam forming, and the interference avoidance capability of a wireless system is improved.

The invention discloses a pilot optimization method for large-scale MIMO channel estimation based on structural compressed sensing. The method comprises the steps of establishing a channel estimation model for a large-scale MIMO-OFDM (Multiple-Input-Multiple-Output-Orthogonal Frequency Division Multiplexing) system when pilots are placed in an overlapping mode; simplifying the channel estimation model for the large-scale MIMO-OFDM system, thereby enabling the channel estimation model to correspond to a structural compressed sensing model; and obtaining an optimum pilot matrix through utilization of a pilot optimization algorithm. Through adoption of the optimum pilot matrix, according to the channel estimation of the large-scale MIMO system based on structural compressed sensing, the mean square errors MSEs of the channel estimation are clearly reduced, and the channel estimation performance is improved.

The invention discloses an industrial equipment fault prediction system based on federated learning and a privacy protection method, and relates to the field of industrial equipment fault prediction. The data acquisition preprocessing module extracts fault features through compressed sensing downsampling, screens and uploads the fault features; the federal learning training module adopts a layered architecture and a dynamic algorithm to schedule a learning rate; the fault prediction and diagnosis module constructs a space-time diagram neural network and fuses a physical model to improve generalization; the privacy protection security communication module performs homomorphic encryption storage and zero-knowledge proof verification update; the knowledge graph construction reasoning module constructs a dynamic graph, locates a fault root cause through causal reasoning, and supports cross-device knowledge migration. By adopting the quantum and federated learning technology, the industrial equipment fault diagnosis accuracy is high, the attack resistance is high, the encryption efficiency is greatly improved, the model training time is shortened, cross-equipment knowledge migration is realized, the operation and maintenance cost is reduced, and the intelligent operation and maintenance development of the industrial equipment is promoted.

The invention provides a training method of a document image tampering detection model and a tampering detection method and device.The training method of the document image tampering detection model comprises the steps that multi-scale visual domain features are extracted from a sample document image, and multi-scale frequency domain compressed sensing features are extracted from frequency domain information; acquiring tampered area edge mask data from the document image; fusing the multi-scale visual domain features and the multi-scale frequency domain compressed sensing features to obtain multi-modal fusion features; performing semi-supervised training on the multi-scale sensing network by taking the multi-scale visual domain feature as a sample feature of a first prediction head, taking the multi-modal fusion feature as a sample feature of a second prediction head, taking a real label or a pseudo label as a sample label and taking joint loss as a loss function to obtain a document image tampering detection model; according to the method provided by the invention, document image tampering pixel-level detection under low labeling cost is realized, and the detection precision of a document image tampering detection model is improved.

The invention relates to the technical field of ground resistance detection, in particular to a single-core cable ground resistance detection method capable of resisting strong power frequency interference. High-frequency signals are injected in a non-contact mode through a coupling type capacitor clamp, background voltage and current signals are collected firstly, and injection frequency far away from power frequency and harmonic waves is selected through Fourier analysis; cancelling background noise by adopting a normalized least mean square adaptive filter to generate an enhanced signal; the signal is down-converted to a base band through digital quadrature demodulation, a Kalman filter estimates the amplitude and the phase in real time, and the state and noise parameters are periodically calibrated through a compressed sensing algorithm; and finally, loop impedance is calculated according to the voltage and current amplitudes and the phase difference to obtain grounding resistance. The method has the advantages of being non-contact, high in safety, strong in anti-interference capability, high in measurement precision and the like, and is suitable for monitoring the grounding state of the cable in a complex electromagnetic environment.

The invention relates to the technical field of safety and environmental protection, in particular to a negative-pressure circulating explosion-control dust removal method suitable for combustible dust places, which comprises the following steps: firstly, sealing a bin body and injecting carbon dioxide-nitrogen inert gas to generate a low-oxygen environment; then, a flow guide angle-adjustable cyclone separator, a dielectric barrier discharge plasma reactor and a waste heat thermo-acoustic transducer are driven to cooperatively capture dust, and sound node amplitude synthesis capture data are obtained through compressed sensing of a microphone; inputting the environment data and the capture data into a reinforcement learning process containing a micro-thermochemistry model, and outputting instructions for adjusting negative pressure pulse, inert flow, sound field intensity and cyclone geometry; in combination with distributed radio frequency imaging data, an adversarial digital twinborn model is generated to predict explosion free energy, and automatic rollback or shutdown is carried out when the explosion free energy exceeds a threshold, so that a safe, efficient and low-energy-consumption closed loop is formed.

The invention provides an automatic processing method for real-time observation data of an ocean station, and belongs to the technical field of ocean observation data processing. A wavelet packet decomposition multi-scale noise separation algorithm is established to distinguish environmental noise and real signals, dual-sensor redundancy configuration is combined with Bayesian inference to identify sensor drift, and a calibration coefficient is updated in real time through a recursive least square method. A one-dimensional time sequence is mapped to a high-dimensional phase space by utilizing a phase space reconstruction algorithm to realize high-precision prediction of a chaotic signal, a hierarchical data storage architecture is established, and a data migration strategy is iteratively optimized through a hierarchical correlation degree function; the technical problem that time synchronization signals are difficult to reconstruct accurately according to asynchronous sampling data of multiple sensors of an ocean station is solved.

The invention discloses a low-rank DCT (Discrete Cosine Transform) compressed sensing algorithm realized based on an FPGA (Field Programmable Gate Array), and relates to the technical field of image data processing, and the compressed sensing algorithm comprises the following specific steps: carrying out blocking processing on an input signal, dynamically adjusting the blocking granularity based on a calculation result of a signal local gradient entropy, and when the gradient entropy is higher than a set threshold value, adopting fine granularity blocking, when the frequency is lower than a threshold value, coarse granularity partitioning is adopted, the partitioning granularity is dynamically adjusted within a preset range, discrete cosine transform is conducted on the partitioned signals, the spectrum entropy of coefficients after transform is calculated, and the number of the reserved low-frequency coefficients is dynamically adjusted according to the magnitude of the spectrum entropy. According to the dynamic blocking strategy based on the gradient entropy, the compressed sensing algorithm can automatically adjust the blocking granularity according to the local features of signals, so that the compression ratio of a complex texture region and a smooth region is improved, the blocking boundary is dynamically judged through the gradient entropy, it is ensured that pixel affiliation of an edge region is more accurate, and the blocking effect is reduced; and complete structure information is reserved for subsequent reconstruction.

The invention discloses a signal acquisition and processing method and system based on a multifunctional radar, and relates to the technical field of signal acquisition and processing, and the method comprises the steps: employing a quantum genetic algorithm to optimize radar transmission waveform parameters, including center frequency, bandwidth and frequency modulation slope, and generating a nonlinear frequency modulation waveform through FPGA hardware; a non-uniform sparse array is adopted to receive a target echo signal, time domain random interval sampling and frequency domain pseudo-random frequency point selection are synchronously implemented, and time-space-frequency three-dimensional compressed sensing observation data are formed; performing trilinear tensor joint sparse reconstruction on time-space-frequency three-dimensional compressed sensing observation data, decomposing a polarization scattering matrix eigenvalue from a reconstructed signal, and calculating a coupling characteristic quantity of eigenvalue entropy and micro-Doppler frequency; inputting the coupling characteristic quantity into a deep reinforcement learning model, and dynamically outputting a constant false alarm detection threshold, a moving target display filter order and a resource allocation weight; and detecting a threshold based on a constant false alarm rate.

The invention discloses a cable accessory nondestructive testing method based on image recognition, and relates to the technical field of data encryption, and the method comprises the steps: extracting a photon signal from an anti-noise coding control template, generating an anti-interference partial discharge light spot image in combination with a compressed sensing reconstruction algorithm, and carrying out the alignment fusion of the anti-interference partial discharge light spot image and LiDAR point cloud data of a cable accessory, generating spatial registration feature data; based on the neural radiation field network, constructing a three-dimensional radiation field model, inputting spatial registration feature data, generating time-space domain deformation field data and partial discharge hot spot distribution data, and generating a defect candidate region by using a spatial weighted fusion algorithm; a deep residual network is used as a framework, a transfer learning strategy is combined, a cable accessory defect model is constructed, a defect candidate area is input, and a cable accessory health state report is generated. According to the method, the three-dimensional radiation field model is constructed by using the neural radiation field network as a framework, accurate positioning and characterization of defects are realized, the detection reliability is improved, and the method is suitable for micro-defect identification.

The invention discloses a multi-medium three-dimensional flaw detection system based on ultrasonic pulse echoes. The multi-medium three-dimensional flaw detection system comprises a synchronous data acquisition module, a temperature compensation module, a multi-physical field simulation optimization module, a signal processing module and a three-dimensional imaging and output module, the synchronous data acquisition module scans an object, records coordinates and synchronously transmits and receives ultrasonic echoes; the temperature compensation module calculates real-time sound velocity according to the water temperature, the real-time sound velocity is used by the multi-physics field simulation module, and the multi-physics field simulation module simulates sound wave propagation and generates a defect-free theoretical waveform as a reference by combining intrinsic parameters of an object and sound wave propagation parameters; the signal processing module carries out processing and feature extraction on actually measured echoes, and compares the actually measured echoes with a reference to identify defect signals; and the three-dimensional imaging module associates the signals with the coordinates, a three-dimensional model is generated by adopting slice reconstruction and a compressed sensing algorithm, and defects are visualized, so that the precision, the efficiency and the environmental adaptability of three-dimensional flaw detection can be improved.

The invention discloses an EIT-based respiratory muscle force and lung compliance closed-loop evaluation system, which is characterized by comprising a multi-frequency EIT injection / sampling unit, a signal processing unit and a signal processing unit, a pressure measurement and synchronization unit; an FPGA clock synchronization triggering unit; an EIT preprocessing unit; a compressed sensing reconstruction unit; a [delta] Z-Pressure fitting unit is adopted; a feature extraction unit; a control decision unit; and the self-adaptive NMES participates in a trend recording unit. According to the method, advanced technologies such as multi-frequency EIT, compressed sensing reconstruction, machine learning fitting, fuzzy + PID mixed control and federated learning are integrated, a new normal form for closed-loop evaluation and optimization of respiratory muscle force and lung compliance is created, and the method has great clinical value for rehabilitation of diseases such as COPD and neuromuscular respiratory failure.

The application discloses a channel estimation method for a millimeter wave frequency division duplex massive MIMO system based on compressed sensing and deep learning, and belongs to the technical field of wireless communication.The technical scheme is as follows: a received pilot signal is processed by using a compressed sensing method to extract a preliminary CSI matrix; a convolutional neural network and a long short-term memory network are combined to extract spatial features and capture time correlation; the preliminary estimated CSI matrix is input into a ConvLSTM layer, and the ConvLSTM network fuses time correlation information to improve the estimation accuracy of the CSI; the same padding, ReLU activation function and filter with appropriate size are used to obtain the same size as the input data; and dimension transformation and inverse normalization are used to obtain the final CSI estimation result.The application has the beneficial effects that the method for channel estimation by using compressed sensing and deep learning aims at reducing the overhead of channel estimation and improving the estimation accuracy in a time-varying channel, reduces the overhead of downlink training and uplink feedback, and improves the channel estimation accuracy in the time-varying channel.

The invention discloses a high-speed high-resolution compressed sensing imaging system and an imaging method. Light of an object is emitted to the polarization splitting prism to be transmitted and reflected after passing through the first-stage imaging module, the transmitted light is reflected back to the polarization splitting prism after being coded and modulated by the spatial light modulator, and then the transmitted light enters the target surface of the camera through the second-stage imaging module to form a coded imaging image; the reflected light is transmitted to the hollow ridge reflecting prism through the quarter-wave plate, is reflected back to the quarter-wave plate through the hollow ridge reflecting prism, then sequentially passes through the quarter-wave plate, the polarization splitting prism and the secondary imaging module and then enters the target surface of the camera to form a common imaging image, and then coded imaging data is matched with common imaging data; and decoding reconstruction is carried out by adopting an alternating direction multiplier method of a point spread function integrated with the first-stage imaging module and the second-stage imaging module. Compared with a traditional imaging mode, the method can break through the imaging frame rate limit of a camera, and has the advantages of high imaging resolution and small data volume.

The invention is suitable for the technical field of radar signal processing, and provides a tomographic SAR super-resolution imaging method based on a structured sparse network, and the method comprises the steps: firstly obtaining multi-channel SAR observation data, constructing multi-channel observation vector MMV data in a data domain, and constructing a multi-pixel signal model corresponding to the MMV data based on a neighborhood pixel elevation consistency hypothesis. The method comprises the following steps: designing a kernel principal component analysis KPCA expansion network model, carrying out dimension reduction and enhancement on MMV data, introducing a norm compressed sensing model on the basis of the MMV data after dimension reduction and enhancement, solving the compressed sensing model by using an ADMM iterative algorithm, expanding the ADMM algorithm into a deep network, reconstructing an elevation spectrum, and obtaining an SAR image super-resolution three-dimensional reconstruction result. According to the method, the super-resolution performance and the solving efficiency of three-dimensional reconstruction can be effectively improved, and high-resolution three-dimensional reconstruction of a large-scale scene can be efficiently realized.

The invention relates to a big model federal splitting privacy protection method based on homomorphic encryption, and belongs to the technical field of big language models and privacy protection. The method comprises the following steps: splitting a large language model into a client side sub-model and a server side sub-model; after the client finishes each round of local training, performing compression processing on an intermediate activation value generated by each client sub-model through a compressed sensing technology; then encrypting and transmitting the compressed activation value by using a homomorphic encryption algorithm, thereby reducing the calculation overhead of the homomorphic encryption algorithm and the communication overhead between the client and the server, and providing privacy protection capability for transmitted sensitive data; the server terminal model is updated by using the gradient of the client, and each client uses local private data to finely adjust the server terminal model, so that the original data of the client is not out of the domain and is available and invisible, and the data privacy security is protected.

A power transmission tower grounding resistance intelligent measurement method based on ground potential field distributed sensing comprises the steps that a ground potential sensing network is arranged around a power transmission tower grounding device, and a ground potential field distributed measurement system with an iron tower as the center is established; a broadband characteristic coding current signal is injected into the grounding system through a weak current injection module; acquiring potential response data of each node of the ground potential sensing network excited by the feature coding current signal by adopting a distributed synchronous acquisition technology, and reconstructing complete ground potential field spatial distribution from sparse node observation data based on a compressed sensing theory through a signal reconstruction algorithm; and inputting the reconstructed complete ground potential field spatial distribution map into a pre-trained map neural network model, and outputting an intelligent identification result and an anomaly diagnosis conclusion of the ground resistance. According to the invention, non-contact, distributed and high-precision grounding resistance measurement under the condition that a grounding lead does not need to be disconnected is realized, and the measurement precision and reliability under a complex working condition are remarkably improved.

The invention provides a radar signal classification method and system based on multi-modal feature adaptive fusion. The method comprises the following steps: performing compressed sensing processing on a time-frequency image by using a pre-constructed sparse sampling matrix to generate observation data; inputting the observation data into a multi-branch feature extraction network, and extracting local texture features and global semantic features through the multi-branch feature extraction network; calculating a global information theory feature tensor based on the time-frequency image, and generating a gating weight matrix based on the global information theory feature tensor; and performing adaptive fusion on the local texture features and the global semantic features by using the gating weight matrix to generate adaptive fusion features, and generating a classification result of the radar signals according to the adaptive fusion features. According to the technical scheme provided by the invention, the data dimension is reduced through compressed sensing, the complementary features are extracted by using the multi-branch network, and the precision and robustness of radar signal classification are effectively improved in combination with an adaptive fusion mechanism guided by an information theory.

The invention relates to the technical field of signal correction, in particular to a nonlinear distortion correction method for a vibration monitoring signal, and provides the following scheme: carrying out broadening and sparse reconstruction on a vibration signal through dispersion Fourier transform and compressed sensing, and extracting a high-frequency local structure; on the basis of phase-space reconstruction and a Lyapunov exponent, nonlinear distortion segments are identified; further performing orbit guiding modeling and degradation path reasoning on the orbit cluster corresponding to the segment, and selecting a reference orbit matched with the structure from the residual linear segments; and constructing an affine mapping relation between orbits and back-projecting the affine mapping relation to a time domain to generate a correction signal. According to the invention, high-precision identification and low-error correction of millisecond-level local nonlinear distortion are realized, and the stability and accuracy of vibration signal analysis are improved.

The invention discloses an image compressed sensing reconstruction method and system based on a recursive diffusion model, and belongs to the technical field of image processing and compressed sensing, and the method comprises the steps: obtaining a to-be-reconstructed original image; performing block compressed sensing sampling on an original image to be reconstructed to obtain a complete observation value; initializing the complete observation value by adopting a pseudo-inverse reprojection operation to obtain an initial reconstructed image; performing complete reconstruction on the initial reconstruction image by using a recursive diffusion model; wherein in the whole image domain, the initial reconstruction image is used as the image estimation of the current iteration, through a lightweight recursion UNet submodule and an operator condition circulation prior submodule in the recursion diffusion model, multi-step iteration reconstruction is carried out, and a final reconstruction image is output. According to the method, a recursive refinement mechanism and stride memory prior are introduced, high-quality image reconstruction is realized under a small number of iteration steps, and the calculation and storage overhead is remarkably reduced.

The invention relates to a transformer abnormal sound source positioning method and system, and belongs to the technical field of power equipment state evaluation, and the method comprises the steps: constructing a Bayesian neural network embedded with a voiceprint physical mechanism, and enabling a sound wave propagation equation to serve as a physical constraint to be embedded into the Bayesian neural network; reconstructing the voiceprint signal by adopting a compressed sensing technology to obtain a reconstructed voiceprint field; designing a multi-task objective function including data fitting, physical constraint and positioning loss, and optimizing data fitting, physical constraint and positioning precision to obtain a trained Bayesian neural network; based on a gradient sound source inversion positioning algorithm and the trained Bayesian neural network, sparse regularization is combined to obtain a prediction result of accurate positioning; and a prediction result is visualized to a three-dimensional model of the transformer, and the position of an abnormal sound source is visually displayed. According to the method, the limitation of a traditional method in a complex environment is overcome, and high-precision and high-robustness transformer abnormal sound source positioning is realized.

The invention relates to the technical field of intelligent monitoring and control of energy storage power stations, in particular to a monitoring method and system for thermal runaway protection of an energy storage power station. The method comprises the following steps: a thermal runaway characteristic signal of the liquid-cooled battery unit is synchronously acquired and processed by the photoacoustic module and the infrared imager to obtain CO concentration time sequence data and temperature cloud picture space data; a sparse feature vector fusing gas concentration and temperature features is obtained through filtering, feature extraction and multi-modal fusion processing of the AI intelligent controller; and performing compressed sensing and risk decision model processing on the sparse feature vector fusing the gas concentration and temperature features to form a thermal runaway early warning signal and a positioning result of the liquid-cooled battery unit. The system comprises a liquid cooling box body, an AI intelligent controller, a cooling liquid storage device, a condenser, a circulating pump, an electric valve, a pressure release valve, an infrared imager, a photoacoustic module and a liquid cooling battery unit. According to the method, rapid extraction and reliable identification of the thermal runaway early-stage features are realized.

The invention provides a geomembrane on-line leakage detection system and detection method, and relates to the technical field of geomembrane detection, the system comprises a signal monitoring module used for arranging a sensing detection network under a geomembrane in an array, carrying out on-line monitoring, and determining a real-time leakage signal of the geomembrane; the sensing and positioning module is used for executing compressed sensing and positioning based on sparse reconstruction according to a plurality of leakage points corresponding to the plurality of leakage signals, and identifying a source sensing matrix; and the leakage early warning module is used for carrying out leakage point early warning based on the source sensing matrix and carrying out leaked area early warning based on the plurality of leakage signals. The technical problems that in the prior art, geomembrane leakage is difficult to accurately recognize and position, response is not timely, and the anti-seepage safety and the management efficiency are reduced are solved. The technical effects of accurately positioning and early warning the leakage points in real time, effectively improving the sensitivity and response speed of geomembrane leakage detection, and improving the anti-seepage safety and management efficiency are achieved.

The invention discloses an Internet of Things flowmeter data compression sensing transmission optimization method and system, particularly relates to the field of low-power-consumption wireless communication networks, and is used for solving the problem of compressed sensing signal distortion caused by sampling time sequence misalignment in an LPWAN bandwidth limited environment. Constructing a local phase reference by extracting a pipeline vibration fundamental frequency component, and receiving a cloud synchronization reference timestamp alignment signal; fundamental frequency and third harmonic components are separated to calculate arrival time difference, and a phase-locked loop is triggered to dynamically adjust the phase of a sampling clock based on a pipeline material threshold value; after the stability of the zero crossing point time variance is verified, compressed sensing sampling is performed on the flow signal by adopting a phase synchronous clock, and an optimized compressed measurement value is generated and transmitted to the cloud through the LPWAN; through closed-loop cooperation of physical characteristics and a communication clock, complete capture of a signal sparse structure is guaranteed under a low-bandwidth condition, and cloud reconstruction precision is significantly improved.

The invention discloses a digital equivalent source sound field reconstruction system based on an underwater complex sound field environment. The digital equivalent source sound field reconstruction system comprises a sound pressure acquisition module which acquires radiation sound pressure data on a measurement point through a hydrophone array; the intelligent analysis module is used for performing shallow sea sound pressure Green function derivation by utilizing a shallow sea Pekeris waveguide model based on the radiation sound pressure data, and constructing an underwater sound field equivalent source equation; the compressed sensing module is used for converting an underwater sound field equivalent source equation into a compressed sensing problem based on the underwater sound field equivalent source equation, and solving a norm for reducing L1 by adopting an algorithm under an accelerated near-end gradient to obtain compressed sensing equivalent source intensity; and the sound field reconstruction module is used for constructing a free field green function, obtaining underwater reconstruction sound pressure by combining the free field green function with the compressed sensing equivalent source intensity, realizing digital equivalent source sound field reconstruction of an underwater complex sound field environment, improving data processing efficiency and reducing requirements on computing resources.

The invention provides a pulmonary artery embolism imaging method, and relates to the field of magnetic resonance imaging. According to the specific implementation scheme, an acquisition window is determined according to electrocardio data and diaphragm navigation data of an adaptive patient; according to the collection window, original frequency domain data are collected in a Cartesian filling mode; performing phase compensation on the original frequency domain data by utilizing pulmonary artery pulse vector field modeling and a dynamic compensation algorithm to obtain proportionality coefficient spatial data; and performing compressed sensing reconstruction on the proportionality coefficient spatial data to generate a three-dimensional pulmonary artery image. The scheme is used for detecting pulmonary artery embolism in the free breathing state, no contrast agent is needed, radiation is avoided, and the universality of applicable people and the inspection safety are improved. According to the technology, by reducing artifact interference and optimizing image quality and scanning efficiency, rapid and accurate diagnosis is achieved, the comfort level of a patient and clinical diagnosis efficiency are remarkably improved, and the technology has important application value and social benefits.

The invention relates to the technical field of spraying control, and discloses a precise spraying control method based on combination of laser scanning and a neural network, and the method comprises the following steps: scanning an environment through a laser radar, dynamically adjusting the scanning harness density based on the real-time speed of equipment, and generating sparse sampling point cloud data; performing compressed sensing reconstruction on the sparse sampling point cloud data, inputting the reconstructed three-dimensional point cloud into a neural network model for target positioning, and outputting a three-dimensional coordinate of a target; calculating a spraying error of the current position of the equipment according to the three-dimensional coordinates; the invention further discloses a precise spraying control device based on combination of laser scanning and the neural network. The precise spraying control device comprises a laser scanning module, a data processing module, a control calculation module, an execution mechanism module and a feedback closed loop module. Through laser radar linkage regulation and control, neural network dynamic allocation and fractional order control, spraying precision improvement, response acceleration and resource optimization are realized.

The invention relates to the field of digital information transmission, in particular to an OTFS channel estimation method based on dynamic sparse Bayesian under a low earth orbit satellite, which comprises the following steps of: converting an information symbol and a pilot symbol into a time delay Doppler domain sending matrix; sending the time delay Doppler domain sending matrix to a receiving end; dividing a pilot signal of a receiving end into a channel vector and a verification pilot signal; modeling a channel vector into a compressed sensing problem based on an OTFS system, and solving the compressed sensing problem by adopting dynamic sparse Bayesian channel estimation to obtain a posterior channel vector; iteratively updating the hyper-parameter of the posterior channel vector to obtain an optimal channel estimation result; performing iterative optimization on a channel estimation result by adopting a verification pilot signal to obtain an estimated channel vector; the OTFS channel estimation method designed by the invention has relatively strong anti-noise and anti-interference capabilities, and the estimation precision is remarkably superior to that of a traditional sparse Bayesian learning method.

The invention discloses a chip defect real-time detection method based on improved YOLOv13, and the method comprises the steps: constructing an industrial chip data set containing a defect-free sample, and carrying out the definition screening and class balance division; a difficult sample perception dynamic enhancement strategy (HSA-DA) is innovatively proposed, a difficult sample is identified based on dispersion, relative size and contrast feature space, and virtual dense cluster enhancement, local super-resolution reconstruction and adaptive contrast stretching are specifically implemented; a network structure is improved, a multi-scale compressed sensing module CS-C3k2 is introduced into a Backbone shallow layer to enhance defect feature extraction, a Sim-DSC3k2 module without parameter SimAM attention is integrated in a Neck deep layer, and a chip defect real-time detection model is constructed based on an improved YOLOv13 network structure and used for chip defect real-time detection. According to the method, 118FPS real-time detection is finally realized under a lightweight parameter quantity (2.45 M), mAP50 is improved to 0.807 (+ 4%), mAP50-95 is improved to 0.573 (+ 7.5%), and the detection robustness of sparse / tiny / low-contrast defects is remarkably optimized.

The invention provides an underwater target detection system and method for an AUV-carried synthetic aperture sonar, and belongs to the technical field of underwater acoustic detection. The method comprises the following steps: transmitting an ultra-wideband sound wave signal to an underwater target, and collecting an echo signal by using receiving array elements which are non-uniformly and sparsely arranged along a track direction; preprocessing the echoes and decomposing the echoes into a plurality of non-overlapping sub-band signals, obtaining an equivalent space response by calculating inter-channel phase differences under each sub-band, and further constructing a joint observation matrix; and high-resolution azimuth scattering intensity distribution is recovered in combination with compressed sensing, and motion compensation is performed by using platform motion data to realize underwater imaging. According to the invention, the dependence of the traditional synthetic aperture technology on long-distance uniform-speed motion is broken through, and high-resolution and high-precision underwater target detection and imaging can still be realized under the condition of high-speed navigation of an AUV (Autonomous Underwater Vehicle).