101 results about "Sparse coefficient" patented technology

The invention discloses a method for establishing a high-voltage circuit breaker fault detection model, and the method comprises the following steps: collecting current, voltage, mechanical response, temperature and other multi-dimensional signals of a circuit breaker under different working conditions, and unifying the signals into standardized time sequence data; a nonlinear dynamic sparse identification method is utilized to establish a dynamic model for describing equipment state evolution, and sparse coefficients reflecting physical change rules are extracted from the dynamic model to serve as health features. And the features are fused with current monitoring data to generate a joint feature input vector, and a health prediction model based on a TabPFN architecture is introduced for training and discrimination. And finally, accurate prediction of the current health state or the potential fault of the circuit breaker is realized, and the model self-adaptive updating capability is realized. According to the method, physical modeling and data analysis are combined, so that the accuracy and interpretability of fault prediction are improved.

The invention relates to the technical field of satellite communication signal processing, and discloses a broadband satellite signal blind demodulation reconstruction method and system based on sparse representation. The method comprises the following steps: carrying out frequency domain transformation on broadband satellite mixed signals to extract pilot frequency features to construct a beam state feature matrix, inputting the matrix into an adversarial network to generate a beam adaptive sparse dictionary, constructing a topological relation graph according to a Doppler frequency shift set, and carrying out collaborative sparse decomposition to obtain a sparse coefficient matrix; and extracting a code rate candidate set to execute parallel sparse reconstruction, determining an actual code rate through self-consistency verification to obtain a complete signal, analyzing a switching instruction, extracting time sequence statistical characteristics, predicting target domain characteristics, generating a switched sparse dictionary, and executing demodulation. The sparse decomposition precision, the code rate blind estimation accuracy, the cooperative processing efficiency and the switching continuity of the broadband satellite mixed signals are improved.

The invention discloses a deep reinforcement learning-based valve flow anomaly detection method. The method comprises the following steps of: obtaining a marked space-time synchronization industrial valve operation data set; generating an enhanced industrial valve image; inputting the enhanced industrial valve image into the improved YOLOv9 detection network, and outputting a candidate micro abnormal region set and corresponding time sequence anchor point information; forming a visual feature set and a time sequence feature set; inputting the visual feature set and the time sequence feature set into an improved coupling convolution sparse coding model to obtain a visual domain reconstruction residual error, a time sequence domain reconstruction residual error and a sparse coefficient consistency deviation; obtaining a normalized abnormal score; and comparing the normalized abnormal score with an abnormal threshold dynamically updated according to the valve flow physical prior constraint, and when the normalized abnormal score exceeds the abnormal threshold, generating micro-abnormal alarm information and an abnormal level. According to the method, false alarm and missing alarm caused by working condition change can be remarkably reduced in practical application, and the reliability of abnormal judgment is improved.

The invention relates to a land surface temperature and emissivity joint inversion method and system and a medium. The inversion method comprises the following steps: acquiring radiation brightness vectors observed by a satellite sensor in a plurality of thermal infrared bands as observation radiation brightness vectors; constructing a standard emissivity spectrum dictionary database as a dictionary matrix; defining a generation rule of the surface emissivity; establishing a forward physical model, and taking an analog value of the observed radiance vector as model output; defining a likelihood function of an observation radiance vector, defining sparsity prior probability distribution for the sparse coefficient vector, and defining uniform prior probability distribution for the surface temperature; joint posterior probability distribution of the surface temperature and the sparse coefficient vector is calculated, and multiple groups of samples are extracted; calculating a posteriori estimation value and an uncertainty interval of the surface temperature; and calculating a posteriori estimation vector of the sparse coefficient vector, and multiplying the posteriori estimation vector with the dictionary matrix to obtain a posteriori estimation value of the surface emissivity. The automation level of remote sensing information extraction and the information output efficiency are improved.

The invention provides a business load prediction method and device, equipment, a storage medium and a product, and relates to the field of financial science and technology or other related fields, and the method comprises the steps: obtaining historical business feature data and a historical business load value; converting the historical business feature data into an original sparse coefficient matrix of a sparse linear equation set; converting a first element value in the original sparse coefficient matrix into a semi-precision format for storage; generating an approximate inverse matrix of the original sparse coefficient matrix based on the first element values stored in the semi-precision format, and storing second element values in the approximate inverse matrix in the semi-precision format; converting the first element value and the second element value in the semi-precision format into a single-precision format; and based on the historical business load value, the first element value of the single-precision format and the second element value of the single-precision format, executing solving operation of the equation set, and determining a predicted value of the business load according to a solution vector obtained by solving. According to the method, the storage space occupied by the matrix data and the data transmission bandwidth pressure can be reduced.

The invention relates to the technical field of data processing and state monitoring, and discloses an intelligent ring main unit fault real-time diagnosis method and system, and the method comprises the steps: constructing a fault feature over-complete dictionary; collecting and preprocessing high-frequency operation data streams of the ring main unit in real time; carrying out online sparse decomposition and manifold mapping based on the dictionary, and converting a high-dimensional signal into a low-dimensional sparse coefficient and manifold features; and a lightweight classification model is used for rapid reasoning, and fault early warning and data uploading are triggered when abnormity is detected. The system comprises a data acquisition module, a preprocessing unit, an edge computing engine, an intelligent diagnosis module and a communication gateway. According to the method, depth dimension reduction and feature enhancement are realized through a manifold sparse coding technology, the bandwidth pressure and the calculation overhead are reduced on the premise of ensuring the fault feature integrity, and the real-time performance and the accuracy of diagnosis are improved.

The invention relates to the technical field of ion beam physical calculation, in particular to an ion beam optical property evaluation method based on a CPU-GPU hybrid parallel stable double-gradient conjugate algorithm. According to the technical scheme, the method comprises the following steps that a CPU-GPU hybrid parallel architecture is initialized, and CPU end and GPU end memory allocation, CUDA related object creation and GPU equipment initialization and parameter setting are completed; constructing a linear equation set, a sparse coefficient matrix A and a right-end vector b, and setting an initial solution vector, convergence precision epsilon and related parameters of the number of iterations; and analyzing the sparseness of the coefficient matrix A. Through combination of three core mechanisms of CPU-GPU cooperative calculation, dynamic load balancing and intelligent preprocessor selection, efficient, stable and universal evaluation of the ion beam optical characteristics is successfully realized, and an excellent solution is provided for solving the solving problem of a large-scale sparse linear equation set in the field of high-performance calculation.

This invention discloses a noise reduction method, a bearing fault diagnosis method, and a system based on dual sparse dictionary adaptive learning. The method performs wavelet decomposition on the signal to be denoised to obtain high-frequency and low-frequency signals, constructing high-frequency and low-frequency matrices. Then, threshold-adaptive DDTF dictionary learning is applied to both the high-frequency and low-frequency matrices. Specifically, an initial dictionary is set and learned to obtain an initial sparse coefficient matrix. An adaptive threshold update is then applied to the initial sparse coefficient matrix, selecting sparse coefficients from the initial sparse coefficient matrix in descending order as dynamic thresholds during the iteration process to update the initial sparse coefficients. After determining the final threshold, the final sparse coefficient matrix is ​​obtained, and the dictionary is updated again to obtain sub-signals. Finally, inverse transformation and matrix rearrangement inverse operations are performed on the obtained sub-signals to obtain the denoised signal. This invention combines wavelet decomposition and DDTF to construct a dual sparse pattern, effectively improving the sparse representation capability of a fixed basis.

The application provides an abnormal audio detection method and a computer device, which can be applied to the field of device detection, and comprises the following steps: obtaining first audio data (including running audio of a to-be-detected device and surrounding environment audio), respectively performing time domain representation and frequency domain representation on the first audio data, then calculating correlation according to obtained first time domain data and first frequency domain data, and obtaining second audio data according to the correlation, extracting target features of the second audio data through a trained first model, obtaining sparse coefficients based on the target features, and judging whether the second audio data is abnormal audio according to the sparse coefficients. The application describes audio change conditions from two dimensions of time domain and frequency domain, simultaneously smoothes a sound scene through correlation comparison of the two dimensions, removes interference, and corrects a starting point and an ending point of current audio data. In addition, key features in the second audio data are identified by using the trained first model, so that various types of abnormal sound can be identified.

The application discloses an aero-engine bearing fault diagnosis method and system, which carries out amplitude normalization preprocessing on the vibration signal of the aero-engine bearing, unifies the signal data magnitude, and eliminates the analysis interference caused by the signal amplitude difference; a convolution sparse coding optimization objective function is built relying on a non-separable learnable sparse regularizer, the limitation of traditional norm and other separable regularization methods is broken, the sparse coefficients are no longer regarded as independent individuals, the structural interaction relationship between the multi-channel sparse coefficients can be fully mined, the interference caused by strong background noise, multi-source vibration coupling and complex transmission path under actual working conditions can be effectively stripped, the weak impact characteristics of the early bearing damage submerged by noise can be accurately separated, the early fault signal extraction effect under complex operating conditions is greatly improved, and the local impact fault characteristics in the aero-engine bearing vibration signal are effectively enhanced, so that the fault diagnosis is completed without any label.

The invention discloses an aircraft shape high-precision aerodynamic design method and device based on a quantum discrete adiabatic algorithm, and relates to the technical field of quantum computing. The method comprises the following steps: constructing a linear system model according to target aerodynamic data based on a computational fluid mechanics method; based on the cyclic shift basis matrix, according to the sparse coefficient matrix and the source item vector, performing block coding by using a linear combination unitary operator technology to obtain a block coding unitary operator; based on a preset iteration operator, constructing a walking operator according to the block coding unitary operator; according to the walking operator, line preparation is carried out through a cyclic calling method, and a quantum solving line is obtained; executing a quantum solving line, and measuring flow field quantum state information; and based on the target aerodynamic data, according to the flow field quantum state information, carrying out high-precision aerodynamic optimization design on the shape of the aircraft by utilizing an aerodynamic integration method. The high-precision aerodynamic design method for the aircraft profile is efficient and accurate based on the quantum discrete adiabatic algorithm.

The invention belongs to the technical field of optical fiber sensing, and relates to an optical time domain reflection monitoring method and device based on forward and reverse signal joint optimization, and the method comprises the steps: obtaining reverse and forward OTDR observation signals of a to-be-measured optical fiber; constructing a reverse OTDR linear signal reconstruction item based on a product of the orthogonal transformation matrix and a to-be-solved first sparse coefficient vector; constructing a forward OTDR linear signal reconstruction item based on the product of the orthogonal transformation matrix and a to-be-solved second sparse coefficient vector; constructing a data fidelity item based on the OTDR observation signal and the linear signal reconstruction item; constructing a joint sparse regular term based on the structural sparsity of the first sparse coefficient vector and the second sparse coefficient vector to be solved; and constructing a spatial constraint regular term based on the continuous smoothness of the reconstructed OTDR linear signal in the space, thereby constructing a joint optimization function and solving the joint optimization function to obtain a first sparse coefficient vector and a second sparse coefficient vector, and further reconstructing the OTDR linear signal to detect the optical fiber to be detected.

The invention relates to the technical field of data cloud storage, in particular to a data backup method and system based on cloud storage, and the method comprises the steps: carrying out the feature collection and processing of to-be-backed-up unstructured data, so as to obtain a joint feature vector; performing matching calculation on the joint feature vector and a preset matrix structure according to a preset tracking algorithm to obtain a sparse coefficient vector; according to the sparse coefficient vector, determining whether the to-be-backed-up unstructured data is duplicated data; and if the unstructured data to be backed up is the non-duplicated data, backing up the corresponding unstructured data to be backed up to the cloud server. Content storage and transmission of repeated data are avoided, cloud storage capacity occupation and bandwidth consumption are reduced from the execution level, and repeated storage and effective metadata retention are both avoided.

The application provides a multimodal large model compression method and device based on sparse codebook quantization, and relates to the technical field of computer vision, wherein the method comprises the following steps: optimizing a visual encoder to make the weight significance distribution of an arbitrary large-scale visual-language model more concentrated; evaluating the influence of each layer weight of the optimized language model on the output through second-order information, and dynamically allocating the code word quantity of each weight group according to the significance; determining the optimal sparse combination from a large-scale codebook by adopting a two-stage strategy of high-level candidate search and low-level subset refinement; and completing model quantization by combining the combination and sparse coefficients, so as to balance the compression and inference performance. The dynamic code word allocation and hierarchical search method based on sparse coding does not need additional training, can adaptively allocate the optimal sparse code word combination, can keep the model expression ability under extremely low bits, and can improve the quantization compression efficiency and inference performance.

This application discloses a method and system for compressing and reconstructing borehole elastic wave signals based on segmented windowing and sparse coding, relating to the field of geophysical signal processing. The method includes segmenting the original borehole elastic wave time-series signal into multiple signal segments; applying a combined Hann window function to each signal segment for windowing and selecting effective signal segments; obtaining the sparse coefficient vector and atom index corresponding to each effective signal segment based on a pre-built dictionary; calculating the average sparse coefficient vector of all effective signal segments to obtain the average sparsity; reconstructing any effective signal segment according to the average sparsity to obtain a reconstructed effective signal segment; and performing inverse weighting processing using the Hann window function corresponding to the reconstructed effective signal segment to obtain the final reconstructed signal. This application achieves efficient compression and high-fidelity reconstruction of borehole elastic wave time-series signals.

This invention relates to the field of wind power equipment fault diagnosis technology, and provides a method for extracting fault features from wind turbine gearboxes based on Gaussian mixture modeling. The method includes: modeling a mathematical model of the wind turbine gearbox observation signal by superimposing impact fault feature vectors and multi-source noise vectors, wherein the fault feature vector is the product of a redundant dictionary D and a sparse coefficient vector; modeling the multi-source noise vector as a Gaussian mixture distribution; constructing an objective function within a Bayesian framework to solve for the sparse coefficient vector in the mathematical model using maximum a posteriori probability estimation; simplifying the objective function; and using the EM algorithm and ADMM algorithm in a joint alternating iterative solution to obtain the optimized sparse coefficient vector; and reconstructing the impact fault feature vector in the wind turbine gearbox observation signal. This method improves the accuracy and robustness of extracting fault features from the observation signal of offshore wind turbine gearboxes.

The invention provides an aircraft electromagnetic scattering data processing method based on scattering source compressed sensing, which relates to the field of aircraft scattering characteristic evaluation, and comprises the following steps: constructing a radar receiving target echo signal equation for measuring the near-field electromagnetic scattering characteristic of the whole aircraft, and obtaining a wave number domain radar receiving echo signal equation through Fourier transform; according to the characteristic of low scattering of an aircraft, it is proved that the equation meets a compressed sensing condition, a classical random Gaussian matrix M is adopted as a sampling matrix, a sparse coefficient vector f solving equation is constructed by adopting L1 norm minimization, and the sparse coefficient vector f is solved based on a classical near-end alternating linearization algorithm. And performing approximate zero processing on a non-zero small quantity in the sparse coefficient vector f by adopting an L1 norm, and finally obtaining the sparse coefficient vector f to obtain aircraft electromagnetic scattering two-dimensional data. According to the invention, the sampling data during the near-field electromagnetic scattering characteristic measurement of the whole aircraft can be reduced, the final data volume is reduced, and the data processing speed is improved.

The invention discloses a local monitoring method and system for the state of printing equipment, and relates to the technical field of operation monitoring, and the method comprises the steps: collecting the multi-dimensional data of the printing equipment, building a time-angle mapping relation, resampling a time domain vibration signal into an angle domain signal, and carrying out the local monitoring of the state of the printing equipment; dividing into a neutral area and an imprinting area according to the physical structure of the roller to construct a state matrix; generating an initial impact feature matrix based on a preset feature element, performing feature screening and reference calibration by using an optimized objective function, and performing sparse decomposition on a state matrix to obtain a sparse coefficient vector; and finally, calculating a steady-state feature vector of each color cell, constructing a dynamic reference, calculating a residual error, and judging the state of the equipment by combining with a sparse coefficient module value change. According to the method, the influence of rotating speed fluctuation is eliminated through angle domain resampling, the interference of the partition processing and sparse decomposition stripping process is utilized, weak fault features are extracted, and high-precision monitoring of the equipment state under the complex working condition is achieved.

A method for detecting abnormal audio, and a computer device, which can be applied to the field of device detection. The method comprises: acquiring first audio data (comprising running audio and ambient audio of a device to be detected), and separately performing time domain representation and frequency domain representation of the first audio data; then calculating a correlation on the basis of obtained first time domain data and first frequency domain data, and obtaining second audio data on the basis of the correlation; extracting a target feature of the second audio data by means of a trained first model; obtaining a sparse coefficient on the basis of the target feature, and on the basis of the sparse coefficient, determining whether the second audio data is abnormal audio. Audio changes are depicted from two dimensions: the time domain and the frequency domain. In addition, a sound scene is smoothed by means of comparing the correlation between the two dimensions, to remove interference, so as to correct a start point and an end point of current audio data. In addition, key features in the second audio data are recognized using the trained first model, so that various abnormal types of sound can be recognized.

The invention discloses a signal feature extraction method of convolution sparse auxiliary filtering, and belongs to the field of signal processing. According to the method, the optimal target atom is obtained by constructing a wavelet matching algorithm. Secondly, convolution penalty is provided, and the sparse coefficient is solved in combination with optimal target atom optimization; in addition, a process parameter optimization selection strategy is provided, and the problem that optimization parameters are difficult to select is solved. And finally, extracting main features of the envelope spectrum, comparing the main features with a theoretical fault feature frequency, and judging a fault type. According to the invention, deep features of signals can be effectively extracted, and efficient fault feature extraction and diagnosis are realized.

The invention belongs to the technical field of intelligent power grids, and particularly discloses a transformer substation terminal box data processing method based on edge computing, which comprises the following steps of: continuously acquiring electric signals from a terminal box and storing the electric signals into an annular buffer area to store high-frequency sampling data; extracting a data window, and decomposing the data window into a steady-state component and a sparse transient component representing terminal micro-contact burrs by means of pre-training over-complete dictionary sparse coding; identifying and slicing candidate glitch event segments according to the energy or peak index of the sparse transient component; inputting the sparse coefficient sequence corresponding to the fragment into a lightweight Transformer model, and judging whether terminal loosening burrs exist or not through a self-attention mechanism; and clustering discrimination results in a preset time window, synthesizing a comprehensive abnormal score, and generating and reporting an early warning if a threshold value is exceeded. Hardware does not need to be newly added, early burrs can be accurately captured, efficient early warning of terminal looseness is achieved, and stability of a power grid is guaranteed.

The application provides a traffic hub vertical field large model training and light deployment method and system, relates to the technical field of machine learning, and comprises the following steps: collecting traffic hub multi-scene service corpus, marking static knowledge, semi-static knowledge, periodic knowledge and dynamic knowledge into four categories according to knowledge timeliness, and constructing a classification data set; performing discrete cosine transform mapping on the weight increment of a pre-trained large language model to a frequency domain space, dividing four frequency band intervals according to the timeliness category, retaining sparse coefficients in each frequency band as trainable parameters, and adopting a differentiated learning rate to fine-tune a teacher model; transferring the teacher model capability to a light student model through progressive three-stage distillation; deploying the student model to a mobile terminal, establishing an end-cloud collaborative mechanism, and updating parameters according to a frequency band level strategy; and responding to user dialogue requests based on the deployed model. The application drives frequency domain sparse distribution according to knowledge timeliness, realizes ultra-low parameter fine-tuning and mobile terminal light deployment.

The embodiment of the invention discloses a random array sound vortex generation method, device and equipment and a storage medium, and belongs to the field of sound vortexes, and the method comprises the steps: building a sound field model from a random array to an observation plane, and determining the sound pressure distribution of ideal sound vortexes; expressing the sound pressure distribution of the ideal sound vortex by using a sparse representation basis, and generating a linear relationship between the sparse representation basis and a driving signal; establishing a relationship between the sparse coefficient and the sound pressure of the observation point by using the sensing matrix; minimizing the error between the actual sound pressure generated by the sparse coefficient through the sensing matrix and the sound pressure distribution as a target, and utilizing L1 norm penalty to make the sparse coefficient sparse to obtain an optimized sparse coefficient; and calculating according to the optimized sparse coefficient to obtain a driving signal vector. The geometric limitation of a traditional uniform circular array is broken through, and the flexibility and robustness of array design are improved. The array configuration and the number of units can be optimized for different scenes.

The present application relates to the technical field of wavelength division multiplexing, and particularly provides a down-wave enhancement method of a reconfigurable optical add-drop multiplexer based on sparse representation, which converts and processes a down-wave mixed optical signal to obtain a digitized electrical signal vector; wherein the down-wave mixed optical signal at least includes a target signal and an adjacent channel interference signal; the electrical signal vector is processed by sparse decomposition based on an overcomplete dictionary to obtain a sparse coefficient vector; wherein the overcomplete dictionary at least includes a target signal atom and an adjacent channel interference atom; a target coefficient corresponding to the target signal atom is identified and extracted from the sparse coefficient vector; and the target coefficient is combined with the target signal atom to obtain an enhanced target signal. The down-wave enhancement method of the reconfigurable optical add-drop multiplexer based on sparse representation can improve the accuracy of extracting the target signal in the ROADM down-wave, effectively suppress adjacent channel interference and amplified spontaneous emission noise, and thus improve performance.

The present application relates to a kind of flip chip vibration signal denoising method, system, equipment and medium, belong to flip chip vibration signal denoising technical field, wherein, method includes: the vibration signal of flip chip is collected;Initial sparse representation of the vibration signal is constructed, and sparse dictionary in the initial sparse representation is constructed according to the vibration signal, sparse representation model is constructed based on the sparse dictionary and the noise variance of vibration signal;The optimal sparse coefficient about the sparse dictionary in the sparse representation model is solved, and simultaneously the optimal sparse dictionary corresponding to the optimal sparse coefficient is obtained;Useless atom and noise atom in the optimal sparse dictionary are eliminated, and the optimal sparse dictionary after denoising;The vibration signal containing noise is reconstructed using the optimal sparse dictionary after denoising and the optimal sparse coefficient corresponding thereto, and the vibration signal after denoising is obtained.The present application can effectively denoise flip chip vibration signal.

The invention discloses an intelligent cloth printing defect monitoring method and system based on machine vision, and relates to the field of image data processing, and the method comprises the steps: collecting a real-time video stream on the surface of a fabric, intercepting single-frame image data, and carrying out the grid segmentation, so as to obtain candidate image blocks; performing feature clustering analysis on sample blocks in the candidate image blocks, and constructing a standard texture feature library; sparse coding operation is carried out on the candidate image blocks based on the standard texture feature library, linear combination weights are solved, and sparse coefficients are obtained; generating a reconstructed image block according to the standard texture feature library and the sparse coefficient; calculating a pixel value difference between the candidate image block and the reconstructed image block to obtain a residual matrix representing a difference degree; and when the energy norm value of the residual matrix is greater than an anomaly judgment threshold value, outputting a marking signal indicating that the corresponding candidate image block has an image defect. By implementing the method and the device, defect detection without a preset template can be realized, and the adaptability of printing defect monitoring is improved.

This invention relates to the field of IEGS probabilistic energy flow technology, and particularly to an IEGS probabilistic energy flow monitoring method based on a sparse arbitrary chaotic polynomial model, comprising: S1, establishing a deterministic energy flow model for an integrated energy system (IEGS); S2, modeling random input variables and configuring corresponding non-Gaussian probability density functions; S3, generating P multidimensional orthogonal aPC basis functions; S4, generating M0 initial collocation points; S5, transforming the problem of solving the sparse aPC coefficient vector A into an l1-l2 norm minimization optimization model; solving for the sparse aPC coefficient vector A under the current collocation point set; and constructing a sparse aPC surrogate model; S6, verification; outputting the final sparse aPC surrogate model; S7, analytically calculating the statistical moments of key output quantities and reconstructing the probability density functions of key output quantities; S8, evaluating the safety margin of IEGS operation. This method can effectively reduce model dimensionality and improve computational efficiency by combining sparse recovery techniques while retaining the good adaptability of the aPC method to non-Gaussian inputs.

The invention provides a hyperspectral image fusion method and device for image adaptive registration, and relates to the technical field of image processing, and the method comprises the steps: taking a first hyperspectral image as a reference, carrying out the feature point extraction and matching based on the first hyperspectral image and a first multispectral image, and obtaining a projection matrix; performing projection transformation processing on the first multispectral image based on the projection matrix to obtain a second multispectral image; determining a spectral response function based on the second multispectral image and the first hyperspectral image; estimating a spectrum dictionary in the first hyperspectral image based on a VCA algorithm; estimating a sparse coefficient matrix in the first multispectral image based on the spectral response function and the spectral dictionary; constructing a second hyperspectral image based on the spectral dictionary and the sparse coefficient matrix; wherein the spatial resolution of the first multispectral image is higher than that of the first hyperspectral image. Therefore, the spatial resolution of the fused image is remarkably improved while the spectral information integrity of the hyperspectral image is kept.

The application belongs to the technical field of optical fiber sensing, and relates to an optical time domain reflection monitoring method and device based on joint optimization of forward and reverse signals. Reverse and forward OTDR observation signals of a to-be-detected optical fiber are acquired. A reverse OTDR linear signal reconstruction term is constructed based on the product of a orthogonal transformation matrix and a first sparse coefficient vector to be solved. A forward OTDR linear signal reconstruction term is constructed based on the product of the orthogonal transformation matrix and a second sparse coefficient vector to be solved. A data fidelity term is constructed based on the OTDR observation signals and the linear signal reconstruction terms. A joint sparse regularization term is constructed based on the structured sparsity of the first sparse coefficient vector and the second sparse coefficient vector to be solved. A spatial constraint regularization term is constructed based on the continuous smoothness of the reconstructed OTDR linear signal in space. Thus, a joint optimization function is constructed and solved to obtain the first sparse coefficient vector and the second sparse coefficient vector, and the OTDR linear signal is reconstructed to detect the to-be-detected optical fiber.

The application discloses a kind of dual-domain constraint hyperspectral image reconstruction methods based on deep learning, it is related to the technical field of artificial intelligence and computational imaging.The application simultaneously imposes constraint in spectral reconstruction domain and sparse coefficient domain, while ensuring that its inherent sparse structure conforms to physical prior, the complement and verification of dual-domain information significantly improve the fidelity of reconstruction result, reduce artifact and noise;Using deep neural network to realize reconstruction, only once forward propagation is needed for new compression measurement value to output reconstruction result, which greatly reduces the computational complexity, improves the reconstruction speed, and has the potential for real-time processing;The sparsity physical prior is explicitly integrated into the network learning goal, which can effectively resist noise interference and reduce the influence of noise on the reconstruction result;Through back propagation algorithm, the dual-head deep neural network is optimized and trained end-to-end, and the best mapping relationship is automatically learned, which greatly reduces the operation difficulty and application threshold of the method.