115 results about "Spatial correlation" patented technology

The application provides a method and system for detecting camouflaged targets by unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs). The system ensures the synchronization of UAVs and UGVs through the global positioning system (GPS), and uses the infrared cameras, binocular cameras, and radar sensors carried by both to efficiently collect data in the target area. The system further preprocesses and analyzes the collected data, and uses improved SSD algorithms and pattern recognition techniques to identify potential camouflaged targets, and performs pattern matching and information fusion to improve the accuracy of the identification results. The SE attention module is introduced innovatively, which optimizes the feature fusion process, enhances the network's spatial correlation capture ability, and enables the system to use global information more effectively, improving the detection accuracy of camouflaged targets. The application solves the problem of poor reliability in traditional methods for detecting camouflaged targets, improves the real-time performance and effectiveness of the cooperative operation of UAVs and UGVs, and significantly improves the accuracy and response speed of camouflaged target identification in battlefield environments. It has important theoretical value and application prospect in various military crisis response.

This invention discloses a method for multi-source data fusion and power prediction of wind turbine units, belonging to the field of power prediction technology. It addresses the technical problems of poor multi-source data fusion and power prediction analysis in existing solutions. By filtering and retaining key features through mutual information, redundant information can be effectively reduced. Dynamic weighting via an attention mechanism makes the fused features more adaptable to real-time operating conditions. Differential weight adjustments for wind speed, terrain, and equipment status ensure high relevance of the fused features even in complex scenarios. Spatial feature extraction captures local spatial correlations, temporal features capture temporal dependencies, and random forests handle nonlinear mappings, solving the problem of insufficient generalization ability of single models. The attention mechanism automatically assigns weights to different time steps and features, avoiding complex mathematical processes such as matrix operations and noise covariance estimation.

The application relates to a water quality change trend rapid prediction method based on multi-source data fusion and physical constraints, and specifically comprises the following steps: step 1: synchronously collecting data such as spectrum information, DO, COD, temperature and pH at key monitoring sites, constructing a water dynamics-water quality coupling equation, and simulating the space-time dynamic distribution of water quality parameters; step 2: outputting a water quality sensitive area through a water dynamics-water quality model, screening sensor layout points by combining information entropy evaluation and spatial clustering, realizing low-cost water quality sensor network deployment through a multi-objective optimization algorithm, and calculating a water body global water quality distribution map by adopting a spatial interpolation method; step 3: analyzing main pollution sources according to the synchronous collection of spectrum information at key monitoring sites, and adopting principal component analysis and an attention mechanism neural network; step 4: based on real-time optical characteristic value-DO data, combining a spatial topological network, a water quality gradient and a cross-region covariance to capture the spatial correlation of water quality parameters between different points, and constructing an optical characteristic value-DO-COD dynamic prediction model; combining pollution tracing and spectrum characteristic correction to correct the COD prediction value, adopting ensemble Kalman filtering to assimilate multi-source data, and improving the model accuracy.

The embodiment of the application provides a high-voltage cable insulation state diagnosis method and system based on a multi-physical field heterogeneous GNN. Applied to the technical field of power equipment state monitoring and fault diagnosis, the method obtains multi-physical field data of a to-be-tested high-voltage cable, and pre-processes the multi-physical field data to obtain standardized feature data; a heterogeneous graph model is constructed based on the physical topology of the to-be-tested high-voltage cable, and the standardized feature data is taken as an initial feature vector of a corresponding node in the heterogeneous graph model; the heterogeneous graph model is input into a heterogeneous graph neural network embedded with physical priori to extract spatial correlation features and time sequence evolution features of the cable insulation state, and to output spatiotemporal fusion features of each node; the spatiotemporal fusion features of each node are input into multiple parallel fully-connected task heads for multi-task joint diagnosis, and corresponding task diagnosis results are output, thereby improving the accuracy, positioning accuracy and engineering usability of cable insulation defect diagnosis.

This invention discloses an urban land use classification method based on the fusion of multi-source data and deep learning, comprising: Step 1: Constructing a spatiotemporal map using Poisson disk sampling and graph neural networks, integrating the spatiotemporal heterogeneity of multi-source data, and dynamically adjusting weights using an attention mechanism; Step 2: Extracting static features from remote sensing and DEM using Swin Transformer, processing dynamic features from social media and mobile phone signaling using LSTM, fusing static and dynamic features through cross-modal attention, and optimizing data matching degree based on Grad-CAM++; Step 3: Combining the person-to-land flow of mobile phone signaling with the spatial correlation of social media activities, predicting trends using LSTM and embedding ecological indicator constraints; Step 4: Constructing a cross-departmental data sharing platform, collaboratively optimizing planning through a unified indicator system, and dynamically adjusting parameters using social media to mine public demands and POI data; Step 5: Introducing ecological footprint and carbon emission indicators to assess sustainability, establishing a monthly update mechanism, and dynamically calibrating model parameters through real-time data streams.

This invention provides a dynamic weighted fusion system and method based on real-time error covariance closed-loop feedback. Addressing the contradiction between underlying physical coupling interference and the need for high-bandwidth dynamic response in non-stationary random environments for large-scale arrays, this invention utilizes a first-order Markov model to quantify the statistical coherence strength of individual observation units. Secondly, it implements physical-layer regularization constraints, eliminating bus capacitance interference and geometric phase deviation through a slope-constrained driving strategy and spatial consistency mapping. In high-frequency sampling environments of 208Hz and above, combined with transient discrimination logic based on full-axis spatial correlation, an analytical solution from a Lagrange optimized functional is used to establish a negative correlation mapping function between weight allocation and the real-time posterior error covariance trace, achieving dynamic gain suppression for non-stationary observation units. This invention achieves optimal fusion of large-scale redundant information while ensuring high-bandwidth real-time response, significantly improving the system's measurement accuracy and robustness.

This invention discloses a fault diagnosis method for aero-engines based on an adaptive spatiotemporal decoupled graph convolutional network, comprising: acquiring and preprocessing multi-channel sensor signals to construct time-series samples; adaptively constructing an adjacency matrix to form a fault signal graph structure representing the spatial correlation of multi-source signals; inputting the fault signal graph structure into a graph convolutional network to extract spatial features; inputting the extracted spatial feature sequence into a bidirectional gated recurrent unit network to obtain temporal features; introducing a global attention mechanism to the temporal features to assign adaptive weights to features at different time steps; decoupling the spatiotemporal features using a variational autoencoder decoupling layer based on a Gaussian mixture distribution; and outputting the fault category of the aero-engine to complete the fault diagnosis. This invention significantly improves the accuracy, robustness, and engineering applicability of aero-engine fault diagnosis under complex operating conditions by jointly modeling the spatiotemporal characteristics of fault signals and effectively decoupling and fusing features.

This invention discloses a radar clutter signal simulation method based on dynamic scene simulation, mainly addressing the problems of low computational efficiency and insufficient flexibility of traditional clutter simulation methods under conditions of large scenes, long time series, and multiple dynamic factors. The implementation scheme includes: constructing a digital simulation scene based on radar signal parameters and dynamic scene information to determine the clutter region to be simulated; discretizing the clutter region into a uniform basic grid in the range-azimuth plane, and then aggregating it into multiple independently processable clutter channels; calculating the amplitude, Doppler frequency, and phase parameters of the center point of each channel in parallel, constructing a two-dimensional convolution kernel reflecting spatial correlation, and using this convolution kernel to efficiently expand the center point parameters to the entire channel, generating a continuously distributed parameter field; and transmitting the parameter fields of each channel to a hardware logic unit to complete signal modulation and synthesis, outputting a high-fidelity radar clutter signal. This invention can achieve rapid, high-fidelity parameter generation from point to field, and significantly improves computational efficiency while strictly maintaining physical accuracy. It can be used for radar system simulation, electronic countermeasures testing, and radar performance evaluation.

Embodiments of the present application disclose a kind of group adaptive integrated control method, device, electronic equipment and storage medium, involve embodied intelligent technical field, wherein, the method includes: according to target scene characteristic, utilize physical modeling and domain randomization technology to generate diversified environment task set, and pre-processing is carried out to ensure data quality, train reinforcement learning coach based on these tasks, combine decoupling and backtracking policy distillation technology to synthesize training sequence, enhance the generalization and adaptive ability of model, adopt long-time memory model structure design base model, integrate multi-agent observation module to capture time sequence and spatial correlation, dynamic noise injection and asynchronous optimization are carried out in training process, the trained field base model is deployed to target scene, form collaborative network, and set fault-tolerant mechanism to deal with agent offline and other sudden situations, continuously collect actual data to optimize model parameters.The present application effectively solves the problems of high customization, weak migration and high maintenance cost in the prior art.

The application belongs to the technical field of measured value transmission systems, and discloses a monitoring data transmission method based on Beidou positioning, which comprises the following steps: acquiring a Beidou timing pulse, geographical coordinates and an ephemeris update sequence identification code, collecting physical state parameter sequences of a monitored object system; generating a measured value fluctuation gradient representing the transient characteristics of the physical state through first-order difference operation; taking the ephemeris update sequence identification code as a synchronization reference, and combining gradient calculation to obtain a dynamic disturbance offset; determining a target sending time according to the basic phase offset mapped by the geographical coordinates and the dynamic disturbance offset; and reporting monitoring data at the target sending time. The application establishes an endogenous coupling mechanism of physical measurement state and measured value transmission timing, utilizes a synchronous ephemeris sequence to drive a terminal to generate a deterministic off-peak displacement, and eliminates the concurrent collision of measurement data caused by spatial correlation physical events.

This invention discloses a wireless channel key generation method and system based on a fluidic antenna. First, a physical layer key generation (PLKG) model based on a fluidic antenna system (FAS) is constructed, completing channel modeling adapted to FAS characteristics. Second, for independent and identically distributed (AID) and spatially correlated channel scenarios, based on the PLKG secret key rate, a key rate optimization problem with transmit power constraints and sparse port activation constraints is further established, and a solution method is proposed to achieve joint optimization of sparse port selection and transmit beamforming vector. Finally, through the entire process of channel feature sequence detection, quantization, information negotiation, and privacy amplification, a symmetric encryption key shared by legitimate parties is generated. This invention achieves low-overhead, high-security, and highly adaptable wireless channel key generation through the dynamic port reconfiguration capability of the fluidic antenna.

The application provides a kind of multi-sensor data credibility evaluation method and system, the technical solution of the application is, based on sensor historical monitoring data, construct the slow time-varying true value estimation model of introducing weighted standard deviation, calculate the time correlation credibility of current measurement value, and determine its proportion weight;Based on the monitoring data of multi-source heterogeneous associated sensor, construct the spatial correlation model based on multi-sensor fluctuation consistency, calculate the spatial correlation credibility of current measurement value, and determine its proportion weight;According to the time and space correlation credibility and its proportion weight, the comprehensive credibility of the measurement data is obtained by linear weighted fusion calculation.This application can solve the technical problems of single dimension, weak anti-interference ability and complex calculation in the prior art.

The application provides a groundwater flow field information simulation method and system, including obtaining simulation basic data, obtaining hydrological event data through the simulation basic data, obtaining spatial correlation characteristics based on the hydrological event data and using clustering operation, and generating a geological hypothesis set; constructing an initial network topology framework and initializing through the simulation basic data and the geological hypothesis set, extracting an initial collaborative sub-network through the initial network topology framework and defining an execution rule; obtaining a correction parameter adjustment scheme set and a chaos parameter adjustment scheme set through the initial network topology framework respectively and generating a final scheme set, simulating based on the final scheme set to generate a deduction simulation result; performing cyclic deduction and obtaining a final simulation result based on a convergence condition, and providing a digital twin simulation method which can be based on and evolve with hydrogeological physical laws.

The application relates to the field of water and electricity management, and discloses a smart water and electricity integrated management system and method based on digital twinning, wherein the smart water and electricity integrated management method based on digital twinning comprises the following steps: uniformly identifying and recombining operation objects distributed in water supply well groups, pump houses, water pools, pipe network nodes, and power stations at all levels, transmission and distribution lines and key load equipment; according to the operation characteristic differences of the water and electricity objects in the multi-layer entity mapping set, continuously running data are split into state sequences that can reflect the changes of emission guarantee loads; potential instability signs of key equipment under current task conditions are dynamically evaluated; when the digital twinning model identifies that there is an abnormal change trend between the water and electricity systems, related water supply units and power supply units are divided into linkage analysis objects according to spatial correlation and scheduling relationship; and the monitoring strategy and the control strategy of the water and electricity system are dynamically adjusted. The application has the advantage of improving the water and electricity guarantee safety during the emission task.

The application relates to the technical field of PVC detection, in particular to a method and system for impurity detection before printing of PVC decals, which comprises the following steps: controlling a multi-waveband spectrum projection unit to project multiple wavelengths of spectrum to the surface of a to-be-detected PVC decal at multiple detection time periods, and controlling a polarization state adjusting assembly to synchronously adjust a polarization angle; acquiring at least one group of polarized spectrum images collected by an ultra-high-definition area array imaging module at the multiple detection time periods; and respectively performing polarization state calibration on the at least one group of polarized spectrum images to obtain polarized spectrum image data after calibration at the multiple detection time periods. Through the multi-waveband spectrum and polarized spectrum technology, combined with texture feature extraction, image segmentation and spatial correlation analysis, the application realizes accurate detection of impurities on the surface of the PVC decal, and especially under the support of high-resolution polarized spectrum image data, potential impurities can be more effectively identified.

Mechanisms are provided for cross-carrier scheduling from a first cell to a second cell in a wireless networking scheme. In one aspect, a method includes receiving, from a base station (BS), a first configuration for scheduling in a first cell having a first subcarrier spacing (SCS), where the first configuration is associated with a first search space in the first cell. The method also includes receiving, from the BS, a second configuration for scheduling in the first cell, where the second configuration is associated with a second search space in a second cell, and where the second cell is associated with a different second SCS. The method further includes determining a number of blind detections (BDs) based on at least one of the first SCS or the second SCS, and monitoring, based on the number of BDs, a downlink control information (DCI) in the first search space and the second search space.

The application provides a blockchain solar eclipse attack detection method based on CNN, and belongs to the technical field of blockchain and network security detection, and comprises the following steps: a collection engine based on a key node deployment of a blockchain P2P network captures bottom-layer communication messages in real time, and pre-processes the bottom-layer communication messages to obtain a two-dimensional gray image; the two-dimensional gray image is input into a pre-trained CNN detection model to capture abnormal texture, edge and spatial correlation features, and output a binary classification result and a binary classification probability value; the binary classification probability value is matched with a preset threshold, and the response level is determined in combination with the attributes of the affected nodes and the proportion of malicious connections; corresponding defense measures are triggered according to the response level, and the defense effect is continuously monitored and fed back to the pre-trained CNN detection model for iterative optimization. The recognition accuracy of the hidden attack is greatly improved, the anti-interference ability in a dynamic environment is significantly enhanced, and millisecond-level real-time detection and protection response are realized.

This invention provides a traffic flow prediction method based on masked autoencoder clustering and adaptive hybrid expert, belonging to the field of traffic flow prediction technology. The method first extracts deep spatiotemporal features of traffic data using a masked autoencoder and then explicitly decomposes nodes into two subsets—high-flow and low-flow—using K-Means clustering. Next, a dual-channel adaptive hybrid expert network is constructed. The high-flow expert uses a fusion pattern library and a self-similarity multi-graph dynamic graph convolutional network to capture complex spatial dependencies, while the low-flow expert uses a simplified spatial module. Both experts share a temporal feature encoder based on Transformer and temporal embedding. Finally, the prediction results of the two experts are fused and output. This invention achieves end-to-end collaborative optimization of explicit data decomposition and prediction tasks, effectively capturing the long-term and dynamic spatial correlations of traffic networks and significantly improving traffic flow prediction accuracy.

The application discloses a power distribution network operation data verification system and method. The system comprises a data acquisition interface, a stream processing module and a data verification module. The data acquisition interface receives operation data streams from monitoring nodes. The stream processing module divides the data streams into continuous batches, sequentially performs time-frequency transformation and spatial correlation transformation on each batch of data, generates a space-time correlation matrix and performs initial anomaly detection, and then releases memory and processes the next batch in a loop. The data verification module combines the detection results of each batch, filters out a high-confidence abnormal position set by counting the occurrence frequency of the same abnormal coordinates in different batches, and then combines a rule library to perform state evaluation and disturbance tracing. The application combines stream processing and cross-batch coordinate consistency verification to realize real-time and high-reliability anomaly detection of power distribution network operation data, effectively suppresses false alarms, and accurately locates complex anomalies from the space-time joint dimension.

This invention discloses a method for detecting micro-UAVs based on shallow feature enhancement and dual-stream boundary aggregation. First, images are acquired and adaptive data augmentation with physical area constraints is performed. Second, an adaptive physical truncation mechanism is introduced into the backbone network to preserve high-fidelity shallow network branches. Then, a dual-stream collaborative aggregation mechanism is constructed, performing orthogonal AC shape splitting along the channel subspace to aggregate high-frequency boundaries and bottom-level topological features respectively. Next, an adaptive spatial correlation pyramid mechanism based on max-pooling is constructed, performing high-dimensional feature energy shaping and background clutter removal through nonlinear morphological anchoring and autocorrelation manifold reconstruction. Finally, a high-resolution detection head with orthogonal decoupling outputs accurate physical coordinates and classification status. This invention effectively solves the problems of high-fidelity extraction of weak features and effective background suppression, significantly improving the confidence and positioning accuracy of micro-UAV target detection, and has important application value and practical significance.

The application discloses a traffic flow prediction method, device, storage medium and product, and belongs to the technical field of data processing. The traffic flow prediction method of the application effectively improves the accuracy and scene adaptability of traffic flow prediction by fusing multi-source spatio-temporal information and constructing an adaptive processing mechanism. Multi-source information fusion provides a complete feature representation containing time rules and spatial correlations for the model, overcoming the limitations of a single data source. The adaptive attention mechanism based on time context can dynamically identify and adapt to traffic change patterns in different time periods such as morning and evening peak hours and holidays, thereby enhancing the model's ability to capture temporal heterogeneity. The spatial lag correlation analysis combined with topological relationships explicitly quantifies the lag effect and influence strength of traffic propagation between upstream and downstream intersections, improving the model's analytical accuracy of spatial coupling relationships. Through the above collaborative processing, more accurate and reliable prediction of future traffic flow is realized.

This invention discloses a method for predicting the remaining service life of a tool based on the spatiotemporal feature fusion network ST-MambaFormer, comprising the following steps: First, multi-source sensor data on the tool wear process and the tool wear amount are collected, and a training sample set is constructed; second, spatial features of the multi-source sensor data during tool monitoring are extracted using iTransformer, and a stacked state-space Mamba model is constructed to obtain temporal features within complex long-term data; then, multi-scale feature and cross-scale deep fusion methods are designed to fuse global features, thereby outputting the prediction result of the remaining service life of the tool; the above process is repeated to complete model training; after training, the monitoring data of a new tool is input into the trained model, and the prediction result of the remaining service life of the tool can be output. This invention can not only efficiently capture long-term temporal dependencies and remain sensitive to key degradation changes, but also efficiently mine the spatial correlation between multiple sensors; based on this, multi-scale feature extraction and cross-scale deep feature fusion are adopted to enhance the global representation capability of tool wear features and improve the accuracy and stability of the prediction model. This provides a practical and valuable solution for tool monitoring and health management in industrial applications.

This invention proposes a reference station network coordinate encryption method based on single-reference-station VRS technology, belonging to the field of satellite navigation and high-precision positioning. The specific process is as follows: The reference station GNSS receiver receives non-differential pseudorange observation data and non-differential carrier phase observation data broadcast by GNSS satellites, forming non-differential observation equations for pseudorange and carrier phase; based on the non-differential pseudorange and carrier phase observation equations, the non-differential error corrections for the pseudorange and carrier phase observations of a single reference station in the reference station network are obtained; based on the principle of spatial correlation of atmospheric errors, during the generation of a single VRS, it is assumed that the atmospheric errors of the actual reference station and the VRS to be generated are consistent. Using the non-differential error corrections of the single reference station, the accurate coordinates of the VRS to be generated, and the satellite position, virtual pseudorange and carrier phase observations are generated to replace the actual reference station, thereby achieving reference station network coordinate encryption; thus protecting the real-time data of the reference station network.

The invention belongs to the technical field of computer vision, and particularly relates to an end-to-end shape recognition method and system based on a complex network. The method comprises the following steps: carrying out dense discretization on the edge of a to-be-identified image to obtain an edge description point set; selecting a series of key points on the obtained edge description point set at equal intervals; extracting local features of a key point surrounding edge description point set by using a shape context; taking each key point as a node, defining edges in the network according to spatial correlation among the key points, and constructing a multi-layer complex network by adopting a dynamic evolution strategy; local features and topological features of the complex network are processed through a graph convolutional network, and robust expression of the shape classification task is obtained; and performing classification through a full connection layer network to obtain a classification result, and realizing end-to-end shape recognition. According to the method, the adaptability and robustness of the method to different shape classification tasks are enhanced.

This invention proposes a method and system for fault early warning and prediction of rotating equipment in thermal power plants based on neural networks. The method includes: acquiring multi-source heterogeneous operating data of rotating equipment in thermal power plants, including vibration signals, temperature signals, rotational speed signals, oil quality signals, and current signals; performing denoising and standardization processing on the operating data to construct a multi-dimensional feature vector containing time-domain, frequency-domain, and time-frequency-domain features; based on the multi-dimensional feature vector, using a convolutional neural network to extract spatial correlation features, modeling long-term dependencies through a bidirectional long short-term memory network, and dynamically strengthening the weight allocation of fault-sensitive features using an attention mechanism to generate a fault identification model; and using the fault identification model to analyze the preprocessed feature data in real time, outputting fault type, remaining life prediction results, and risk level classification, wherein the risk level classification is generated based on a joint decision of fault severity and remaining life.

This invention discloses a physical layer key generation method and system based on a polarization-reconfigurable antenna, belonging to the field of wireless communication technology. The method includes: establishing a polarization channel model jointly considering polarization correlation, spatial correlation, and channel depolarization effects, and deriving a closed-form expression for the key generation rate; constructing a joint optimization problem of beamforming vector and PRA phase shift vector under base station transmit power constraints and antenna phase modulation constraints, with the goal of maximizing the key generation rate; solving the problem using an alternating optimization method to obtain the optimal beamforming solution based on the Rayleigh quotient and the analytical closed-form solution for polarization phase modulation jointly determined by the polarization channel correlation coefficient, inverse cross-polarization discrimination, and receiver polarization vector. This invention requires no additional RF link, has low hardware complexity, effectively mitigates performance loss caused by polarization mismatch, and significantly improves key generation efficiency and system security in complex polarization environments.

The application relates to the technical field of personnel position determination, in particular to a key personnel position determination method, device and equipment, wherein the accuracy of historical personnel position determination, a historical average time attenuation factor and an average spatial correlation degree are used to dynamically and adaptively determine the weight corresponding to the perception data type. Based on the determined weight corresponding to the perception data type, the time attenuation factor can improve the time sensitivity of the personnel position determination, and the address confidence degree is fused to finally accurately determine the comprehensive score of the address, and the key personnel position is determined based on the comprehensive score of the address, so that the working efficiency can be improved, the personnel cost can be reduced, and the response speed can be improved.

A double-layer robust optimization method for section location in resilient distribution network considering failure probability and reinforcement strategy is proposed, including: establishing a failure probability-aware chance constraint model to quantify the failure probability and its spatial correlation of each section and generate a high-confidence failure scenario set; establishing a reinforcement strategy optimization model to select key sections for pre-disaster reinforcement by genetic algorithm and change the underlying failure risk distribution; establishing a basic model for failure section location under information distortion; combining the chance constraint model, the reinforcement strategy optimization model and the basic model for failure section location, a double-layer robust optimization model for "defense and location" coordination is constructed, and a hybrid intelligent optimization method is used to solve it. Through the coordination of probability awareness, spatial correlation modeling and reinforcement strategy, the method significantly improves the positioning accuracy and robustness of the distribution network under multiple failure and information distortion scenarios, providing a coordinated optimization decision basis for pre-disaster defense and disaster location in resilient distribution network.

The application discloses a kind of time-frequency-space three-level fusion industrial multi-sensor signal anomaly detection method, it is related to signal processing and abnormal detection technical field, first, by local and global context, frequency domain amplitude spectrum and spatial correlation mode Local and global mode four-order tensor are constructed, realize data level time-frequency-space fusion;Second, orthogonal tensor decomposition and bidirectional large kernel convolution network are used to extract time domain, frequency domain and spatial features, and are spliced and fused, realize feature level time-frequency-space fusion;Finally, from variable internal context and variable intercoupling double perspective Local-global contrast difference is calculated, and is weighted fusion, generate the final abnormal score of each timestamp, realize decision level time-frequency-space fusion;According to final abnormal score, whether corresponding timestamp is abnormal is judged.The abnormal detection method of the application, by data-feature-decision time-frequency-space fusion and local-global contrast difference learning, realize high-precision, high-robustness timestamp-level anomaly detection.

The invention discloses a submarine cable state remote wireless telemetering system and method based on LoRa, and belongs to the technical field of cable laying state monitoring, and the system comprises a plurality of state monitoring nodes arranged on a submarine cable, a relay node and a shipborne monitoring center. The analysis computer adaptively adopts different compensation strategies according to the monitored packet loss and missing proportion; during light packet loss, trend filling is executed by using the change rate of effective nodes and spatial correlation; during moderate packet loss, confidence weight dynamically attenuating along with interruption time is introduced into the reconstruction model; and during severe packet loss, extracting feature sub-blocks of the spatial physical association constraint matrix to execute order reduction calculation, and recovering data in combination with a spatial interpolation method. According to the method, the problem of data missing caused by unstable wireless communication under the extreme sea condition is solved through a hierarchical compensation mechanism, it is ensured that the state distribution of the whole submarine cable can be stably and accurately reconstructed and early warning can be output under the working conditions of different degrees of packet loss, and the robustness and construction safety of a submarine cable monitoring system are remarkably improved.