99 results about "Computational topology" patented technology

The invention discloses an intelligent electric equipment monitoring and optimizing method, and relates to the technical field of electric power, and the method comprises the steps: collecting high-dimensional voltage-current time sequence data and transient event marks, calculating topological invariant features, inputting the topological invariant features to a lightweight neural network model, and recognizing the features of all electric equipment; performing abnormal attribution and anti-fact energy efficiency prediction by using causal reasoning and dynamic regularization regression based on the identified characteristics of each electric device, and constructing a multi-objective optimization function through the abnormal attribution and anti-fact energy efficiency prediction; and based on the multi-objective optimization function, generating an equipment operation scheduling strategy through a deep reinforcement learning agent, based on the operation scheduling strategy, sending a control instruction to the electric equipment, and collecting an operation result feedback in real time for optimization and updating. According to the method, through fusion of topological features, causal reasoning and safety reinforcement learning, the precision, robustness and safety of monitoring and optimization of the intelligent electric equipment are improved.

The invention belongs to the technical field of pipe network monitoring control, and particularly relates to an urban pipe network leakage detection method and system.The urban pipe network leakage detection method includes the steps that firstly, data are collected at key nodes and pipe sections of a pipe network through a data collection module, and a four-dimensional original data set with spatial positioning attributes is formed; the preprocessing module adopts an improved Kalman filtering algorithm containing a pipe network material attenuation coefficient to reduce noise, and outputs high-quality data; the space-time fusion module is combined with GIS pipe network topological data, the topological weight is calculated through a node degree centrality algorithm, and a time attenuation mechanism and adjacent node features are fused to generate a comprehensive feature matrix; the dynamic judgment module judges suspected leakage based on a multi-parameter weighting model and a sliding window dynamic threshold value; the positioning engine module positions a leakage point through a two-factor algorithm of signal time difference and improved hydraulic model residual error; and finally, the linkage control module starts secondary acoustic verification, valve control, early warning and maintenance scheduling instructions are automatically generated after confirmation, and closed-loop management from detection to disposal is achieved.

The invention relates to the technical field of water conservancy project structure health monitoring and intelligent sensing, in particular to a water conservancy project online monitoring and abnormity diagnosis system based on edge calculation, which comprises a data acquisition module used for acquiring current water level data, current temperature data and actually measured structure response data; the reference reconstruction module is used for generating an ideal state vector; the fault simulation module is used for generating a theoretical damaged simulation state vector; the topology judgment module is used for calculating a first difference value between the actual measurement structure response data and the ideal state vector, calculating a second difference value between the theoretical damaged simulation state vector and the ideal state vector, and calculating the topology similarity based on the first difference value and the second difference value; the risk assessment module is used for generating a structure safety risk index in combination with the fault strength scalar and the topological similarity; according to the method, the problems of data transmission lag and low online calculation efficiency in a traditional centralized processing mode are effectively solved, and the real-time requirement of an engineering site is met.

The invention discloses a three-dimensional CAD model feature recognition method. The method comprises the following steps: analyzing boundary representation data of a CAD model and constructing a topological graph structure; the feature complexity is quantified by calculating a topological entropy value; in combination with a predefined process semantic rule base and a graph neural network model, realizing association identification of geometric features and processing semantics; and outputting the parameterized feature tree and supporting CAM system integration. The technical problems that a traditional feature recognition method is poor in adaptability to a complex topological structure and lacks process semantic association are solved, and the recognition accuracy and the automation degree are remarkably improved.

The invention discloses an intelligent evaluation system and intervention method for mental health of teenagers, and the method comprises the following steps: collecting multi-mode psychological signal data of the teenagers, and carrying out the preprocessing; establishing a cognitive energy field, and generating a cognitive energy time sequence curve; performing spectrum resonance analysis, and combining the spectrum resonance analysis with a cognitive energy time sequence curve; constructing a psychological topological graph, and calculating a topological complexity index; when the topology complexity index is continuously abnormal or the cognitive energy disturbance amplitude exceeds a set threshold value, a psychological risk early warning result is generated; after the psychological unstable state is detected, a cognitive energy feedback parameter set is generated, and cognitive energy balance intervention is executed; and after the intervention is finished, calculating an energy balance improvement rate, and when the energy balance improvement rate exceeds an energy balance threshold, judging that the intervention is finished. According to the method, multi-mode psychological signals and cognitive energy field modeling are fused, intelligent evaluation and dynamic intervention of the psychological state of the teenagers are achieved, and the method has the advantages of being high in accuracy, high in interpretability and closed in intervention loop.

The invention discloses a semi-supervised medical image segmentation method based on causal uncertainty decomposition, and belongs to the field of medical image processing and artificial intelligence. According to the method, a segmentation model of teacher and student architectures is constructed, and a causal uncertainty decomposition module, a self-adaptive consistency learning module and a topology perception consistency loss module are integrated. The total uncertainty is decomposed into cognitive uncertainty and random uncertainty, so that targeted processing is realized; a dual-path weight fusion and differential modulation strategy is adopted to realize pixel-level adaptive learning; and a Betti number is introduced to calculate a topological distance, so that the integrity of an anatomical structure is kept. According to the method, under the condition that only 5%-20% of annotation data is used, the Dice coefficient on multiple medical image data sets is increased by 3.2%-4.8%, the segmentation precision and the boundary positioning accuracy are remarkably improved, the segmentation problem under the condition that medical image annotation is scarce is effectively solved, and the method has important clinical application value.

The invention provides an energy storage device damage diagnosis and interaction system based on penetration vision and a large model, and relates to the technical field of energy storage device damage diagnos.The energy storage device damage diagnosis and interaction system comprises a penetration type 3D sensing module, a periodic topology analysis module, a feature mapping and retrieval module and a large model reasoning and interaction module, and the energy storage device is scanned and reconstructed; the method comprises the following steps: constructing a time-space decoupling periodic Transform network, introducing a periodic mask matrix to force the network to pay attention to a repeatability rule of an internal structure of a battery, and identifying internal tiny deformation and structural damage by calculating topological consistency under the condition that a large amount of negative sample training is not needed; visual defect features are mapped into text embedding by utilizing a feature projection technology, a diagnostic report containing physical cause analysis and maintenance suggestions is generated by combining a retrieval enhancement generation technology and a large language model, and a user is supported to perform interactive questions and answers in a natural language. The method can solve the problems that in the prior art, three-dimensional deformation is difficult to quantify, small samples are difficult to train, and intelligent decision-making ability is lacked.

The invention discloses a hydroelectric equipment anomaly detection method based on physical mechanism guidance and sequential topological entropy fluctuation characteristics, and belongs to the technical field of hydroelectric equipment monitoring and fault diagnosis. The method comprises the steps that multi-source sensor data of hydroelectric equipment is collected and preprocessed; constructing a physical weighted distance function in combination with an equipment physical mechanism, and embedding time sequence data into a high-dimensional point cloud space; extracting persistent homology features through a sliding window, generating a persistent graph sequence and calculating topological feature indexes; a persistence graph entropy fluctuation index is provided, and anomaly detection is realized by quantifying time sequence fluctuation of topological entropy; and finally, visual output and an alarm mechanism are combined to assist diagnosis. According to the method, equipment physical characteristics and topological data analysis are fused, the problems that a traditional method is insufficient in nonlinear system modeling, insensitive to dynamic evolution and the like are solved, the early warning capacity and detection precision of early faults are improved, and the method is suitable for anomaly detection application of core equipment such as a water turbine and a generator.

The invention provides a historical load analysis-based power supply path dynamic generation and evaluation method, which comprises the following steps of: acquiring historical load data of a commercial district and topological structure data of a power distribution network, extracting load space distribution characteristics, and determining a dynamic change trend of load space distribution by adopting a clustering algorithm; performing time sequence analysis on the offset distance of the load center and the change rate of the power utilization density, and identifying the change period of the load in the commercial district and the sudden characteristics of sudden increase or sudden decrease of the load; key section distribution is identified according to the section identification standard, the load matching degree of the topological structure of the power distribution network is evaluated according to the key section distribution, and the load distribution balance and the line bearing capacity between topological nodes are calculated; and identifying weak nodes and areas with unbalanced load distribution according to the load distribution balance among the topological nodes and the line bearing capacity, and adjusting the line layout according to the identified weak nodes.

The invention relates to the technical field of power system automation and artificial intelligence, and particularly discloses a power network risk alarm event analysis method, which comprises the following steps of: firstly, carrying out standardization processing and space-time slicing on a multi-source heterogeneous original alarm flow, and constructing a dynamic fault propagation graph in combination with power grid physical topology; then, multi-dimensional indexes such as topology centrality, time sequence attenuation and semantic rare degree are comprehensively calculated, and the causal significance of each alarm node is quantified; on this basis, adaptive sparse sampling based on maximum interpretation gain is implemented, and high-score core events are precisely preserved and public precursor nodes with global collective power are forcibly recalled by evaluating common-mode interpretation utility and space-time coupling strength of nodes to downstream alarms on the premise of strictly complying with the limitation of context tokens of a large model. And finally, inputting the simplified high-value evidence chain into a large language model to carry out thinking chain reasoning, thereby realizing accurate positioning of complex fault root causes and automatic generation of a disposal strategy.

The invention discloses a temperature control system and method of a micro-fluidic chip, and relates to the technical field of micro-fluidic temperature control, and the method comprises the following steps: in the manufacturing process of the micro-fluidic chip, embedding topological insulator nanowires into a substrate in a directional arrangement mode to form a pre-locking array, and meanwhile, generating an anisotropic heat conduction network under the continuous action of a sound-magnetic field, a topological vortex invariant is calculated, and a heat conduction weight coefficient is generated; arranging a fluorescent layer on the surface of a micro-reaction cavity of the anisotropic heat conduction network, collecting a fluorescence lifetime attenuation curve in real time, and calculating a transient temperature value and a temperature change rate according to a heat conduction constraint variation model and a heat conduction weight system; nonlinear feed-forward compensation power is calculated according to the transient temperature value and the temperature change rate, feedback compensation power is calculated in combination with a deviation field of the target temperature and the real-time temperature, and a compensation power instruction is generated; according to the method, the thermal pilot frequency domain filtering function is constructed through the compensation power instruction, so that the uniformity and response speed of thermal response are improved.

A computer-implemented method includes loading a simplistic network-on-chip (NoC) topology that is fully routed, and performing reinforcement learning on the NoC topology to identify a sequence of topology transformations that will produce a more optimal NoC topology. Performing the reinforcement learning includes running a plurality of training sessions. Running each training session includes using a machine learning model to apply a set of transformations to the NoC topology according to a policy, computing a cost of the NoC topology after the set of transformations has been applied, and updating the policy in response to the cost.

The invention discloses a defect hidden danger data diagnosis method based on a distribution network SVG single line diagram. The method comprises the steps that basic analysis is conducted on the input SVG single line diagram, basic geometric figure elements are extracted and normalized, a topological connection relation is preprocessed, and a multi-level index structure is constructed; a dynamic topology path reasoning mechanism is constructed, and reasoning and self-adaptive optimization of the topological relation of the electrical equipment are realized through path consistency calculation, topological entropy minimization and a dynamic topology updating strategy; a self-adaptive implicit relational knowledge graph is constructed, and semantic association between the electrical equipment is established through explicit topological relation fusion, implicit function dependence reasoning and dynamic semantic embedding optimization; and constructing a multi-modal time sequence feature vector, calculating the hidden danger risk of the equipment by adopting a space-time defect probability estimation model, and giving an alarm through a dynamic risk early warning strategy to realize prediction and early warning of the hidden danger of the defects of the electrical equipment. According to the method, the problems of semantic understanding, topological reasoning, state inference and defect prediction of the SVG single line diagram are effectively solved.

The invention relates to the technical field of short message transmission optimization, and discloses an intelligent short message resending system and method based on receipt state recognition. A communication data acquisition module of the system obtains historical short message transmission records in a preset time period of a target communication area and generates a communication state data set; a feature extraction storage module performs receipt feature extraction on the data set to obtain a historical receipt feature set and establish a receipt state change matrix; the topology modeling module constructs a terminal state atlas according to the matrix, and calculates the topology aggregation degree of the historical short message failure events to determine an abnormal characteristic index set; a real-time analysis module extracts real-time receipt features of the current short message transmission data to generate real-time state vectors, and calculates a topology correlation degree with the abnormal feature index set in a terminal state graph to generate real-time risk factors; and the decision generation module generates a short message resending strategy instruction according to the real-time risk factor, the historical receipt feature set and the receipt state change matrix.

The invention discloses a dynamic calculation topology optimization method based on an Flink SQL, relates to the field of big data and real-time calculation, and aims to re-realize self-defined calculation topology arrangement and complete dynamic compression of different SQL calculation topologies by matching a native Flink SQL execution plan. The invention discloses an Flink SQL (Structured Query Language) engine-based dynamic combination, adjustment and removal method, and aims at constructing an innovative optimization method based on an Flink SQL engine, and realizing dynamic combination, adjustment and removal of multiple types of SQL calculation operators including Project, Filter, Aggregate, Window and the like in a unified calculation topology aiming at typical calculation modes including data extraction, window aggregation and the like which are covered by a streaming SQL calculation scene based on the same data source, so as to realize dynamic combination, adjustment and removal of multiple types of SQL calculation operators including Project, Filter, Aggregate, Window and the like. Sharing and multiplexing of computing resources are achieved through an intelligent resource scheduling mechanism, dynamic evolution of real-time SQL logic is effectively supported, and the computing efficiency and the resource utilization rate are remarkably improved.

The invention discloses a power distribution network fault data preprocessing method and system based on multi-source information fusion, and relates to the technical field of power systems, and the method comprises the steps: constructing a dynamic topology manifold based on the physical connection relation of power distribution network equipment and multi-source data; manifold deformation caused by line switching or sudden load change is responded, and abnormal data are recognized in combination with the manifold curvature and the equipment connection path distance; encoding the abnormal data, the meteorological data and the equipment state data into quantum states, and generating entanglement feature vectors through quantum gate operation; performing chaos sequence modulation on the basic features to generate modulation features; performing quantum fusion transformation on the modulation feature and the entanglement feature vector to generate an enhanced feature; calculating topological space correlation, chaos behavior similarity and quantum correlation based on the enhanced features, and constructing a multi-dimensional fusion weight; and performing weighted fusion on the enhanced features according to the multi-dimensional fusion weight, and outputting a preprocessed fault feature vector.

The invention relates to the field of computer vision, in particular to a bovine embryo quality evaluation method and system based on a time sequence microscopic image, and the method comprises the steps: collecting a microscopic image and a timestamp, carrying out the embryo segmentation and inter-frame registration of the image, and obtaining the frame quality weight, morphological characteristics, image block embedding and topological data; respectively calculating a topological distance and an optimal transmission distance, determining a change point generation event sequence in combination with a frame quality weight, and aligning and outputting a penalty term based on a time Petri network; inferring a stage sequence and a quality posteriori based on the event sequence and a penalty term, and constructing a multi-instance set based on an event time window to obtain a multi-instance learning score; and fusing quality posteriori and multi-instance learning score to obtain an evaluation result, calculating a conflict coefficient, and relocating an event time window when a threshold value is exceeded to update the evaluation result. According to the method, rating consistency, noise immunity and traceability are improved.

The invention relates to a GAT-based industrial user voltage sag sensitivity degree evaluation method and equipment, and the method comprises the steps: obtaining the characteristics, types and equipment connection relations of sensitive equipment influenced by voltage sag, constructing a topological graph, inputting the topological graph into a trained graph attention neural network, outputting a voltage sag sensitivity evaluation result, and carrying out the evaluation of the sensitivity degree of the industrial user voltage sag. Wherein the graph attention neural network is constructed by embedding a GAL layer in a message passing graph neural network, the GAL layer calculates an attention coefficient between nodes in a topological graph by using an attention mechanism, weighted aggregation is performed by using an update function, iterative update of node features is completed, and a voltage sag sensitivity evaluation result is finally mapped and output. Compared with the prior art, the method has the advantages of improving the evaluation accuracy and the like.

The application discloses a kind of metering acquisition full scene automation simulation detection method and system, this method includes: selecting detection target for the equipment to be measured and associating district topology template, impedance parameter and load strategy;Utilize the technology of multiple bypass wiring to switch physical line connection automatically through network relay, reconstruct target district topology structure;Through compensation algorithm adjustment adjustable impedance equipment output, make the line impedance of the equipment to be measured consistent with target impedance value;Controllable load device is automatically controlled and generates load timing;Issue instruction to the equipment to be measured and collect response data;Response data is stored in association with target district topology structure, line impedance and load timing, and the matching degree of topological recognition result and district topology template and impedance deviation is calculated based on graph isomorphism algorithm, and detection report is generated.The application has the advantages that full scene automation of metering acquisition equipment detection is realized, and detection efficiency, accuracy and repeatability are improved.

This invention relates to the field of architectural planning technology, specifically to a method and system for urban and rural planning based on digital twins. The method includes acquiring building information models and geographic information system (GIS) data of the area to be planned; constructing a three-dimensional digital twin basic scene through spatiotemporal registration and semantic alignment; identifying planning entities and multidimensional constraints to construct a topological hypergraph model; calculating topological centrality and constraint coupling weights to generate a non-uniform computational potential energy field; constructing a hierarchical scheduling strategy based on the potential energy field gradient; loading a LOD3 level fine model to extract topological parameters for high-potential-energy sensitive areas; loading a LOD1 level block model to extract filling data for low-potential-energy background areas; inputting a generative layout algorithm to generate intermediate data; detecting and repairing geometric overlaps; and outputting a planning scheme. This invention, by constructing a non-uniform computational potential energy field and a hierarchical scheduling strategy, achieves on-demand loading and differentiated processing of large-scale scene data, effectively solving the computational bottleneck problem of processing massive heterogeneous data.

Devices, systems, methods, and processes for in-network computing using modular switch architecture are described herein. Endpoint devices generate data chunks and forward them to a network, comprising spine and leaf switches, for data reduction. Leaf switches act as conduits and forward the data chunks to a spine switch. The spine switch includes various line cards (e.g., one for each leaf switch) and a fabric element. The line cards may execute a stage of data reduction on the received data chunks or may forward the received data chunks directly to the fabric element. The fabric element executes a data reduction operation on the data received from the line cards and obtains a reduced output which is forwarded to the endpoint devices via the line cards and the leaf switches. Thus, a single-tier in-network computing topology is implemented to execute data reduction in a cost-effective, simple, and efficient manner.

The present application relates to the technical field of cell image analysis and medical artificial intelligence, and discloses a cell segmentation and adaptive cascade reasoning method and system based on prior frame guidance. The method comprises: acquiring a multi-modal cell image with a detection frame; performing adaptive histogram equalization preprocessing on the image and executing global preliminary screening segmentation; calculating a multi-dimensional weighted score of the mask and the detection frame, if the score is lower than a first matching threshold, extracting a local image block to dynamically adjust a flow field threshold and an estimated diameter for cascade retry; if the score after retrying is lower than a second threshold, triggering a label consistency protection mechanism to discard or back up poor samples; finally, performing connected domain purification on the retained mask, and calculating a topological solidity, and performing convex hull reconstruction on the mask with low solidity. Through two-stage reasoning, double-track quality filtering and topological constraint, the present application effectively improves the segmentation effect of dense and irregular cells and improves the generalization performance of downstream analysis.

The invention discloses an Internet topology enhanced detection method based on decision optimization, and belongs to the field of Internet topology detection. According to the method, firstly, a topology detection knowledge graph is constructed to represent prior knowledge related to topology detection, then a heuristic rule and a strategy network are constructed based on the knowledge graph to select an optimal topology detection decision, and then a self-adaptive reinforcement learning training optimization strategy network is rewarded by calculating the gain of topology detection to update a topology detection strategy. And finally, triggering a detection strategy aiming at multiple paths by identifying the multiple paths and judging the path stability. Through heuristic screening, enhanced strategy decision and iterative updating of detection feedback, the detection process can automatically explore a high-value area, reduce repeated detection, improve the topology coverage rate and effectiveness and ensure the authenticity, integrity and dynamicity of Internet topology detection in a large-scale Internet environment with strong dynamicity and strong isomerism.

The invention discloses a multi-modal lung cancer risk dynamic assessment system and method, and belongs to the technical field of medical data analys.The multi-modal lung cancer risk dynamic assessment system comprises a data acquisition module, a topological representation module, a self-adaptive attention gating module, a risk assessment module and a feedback optimization module; the method comprises the following steps: constructing a family history topological graph, a gene mutation topological graph and an environmental exposure topological graph, extracting multi-level topological features, generating a continuous bar code and a Betti number sequence based on continuous homology analysis, calculating an interaction coefficient, and triggering a cross-modal risk collaborative amplification mechanism when a gene mutation and environmental toxin coexistence mode is detected. The system is further provided with a three-level collaborative amplification mechanism, a risk decomposition report containing gene risks, environment risks, family risks and collaborative amplification factors is generated, dynamic parameter optimization is achieved by calculating the topological entropy change rate, deep fusion of multi-modal data is achieved, and the risk factor interaction effect is accurately quantified.

The application discloses a kind of based on topological analysis's graphic image splicing fusion method and system, belong to image processing field, the method includes: obtaining first image and second image, generate first binary image and second binary image;Respectively topological feature analysis is carried out, and the topological skeleton of image foreground area is extracted and topological invariants is calculated, and initial matching point pair set is generated;After alignment, the first image is generated;Image splicing seam is solved in the cost matrix by dynamic programming algorithm, and the target image after splicing fusion is output.The present application prevents tearing and overlapping distortion phenomenon in image alignment process.

The application relates to a spatial transcriptome spatial domain identification method and system based on double contrast learning. The method comprises the following steps: performing Gaussian smoothing processing on gene expression data to obtain smoothed gene expression data; constructing an initial adjacency matrix based on spatial coordinate data, performing edge pruning to obtain an optimized adjacency matrix; inputting the smoothed gene expression data and the optimized adjacency matrix into a graph convolution network encoder to obtain low-dimensional latent features and reconstruction features, and calculating a reconstruction loss; inputting the low-dimensional latent features into a multi-branch multi-layer perception fusion module to obtain enhanced features; constructing a regional contrast learning task, calculating a regional contrast loss, constructing a topological contrast learning task, and calculating a topological contrast loss; optimizing model parameters through a total loss, using an optimized model to perform spatial domain identification, and obtaining a spatial domain identification result. The spatial domain identification accuracy can be improved and the cost can be reduced.

The invention discloses a mammary gland endoscope preoperative anatomical variation identification method and system, and relates to the technical field of computer data processing. The method comprises the following steps: S1, acquiring heterogeneous source data, and performing spatial registration and data standardization fusion to generate a multi-dimensional fusion data set; s2, performing basic topological feature extraction based on the multi-dimensional fusion data set, and calculating edge weight information to construct a target topological graph; s3, matching and comparing the target topological graph with a preset standard topological database, and calculating topological compliance parameters based on comparison differences; s4, identifying a parameter abnormal region according to the spatial distribution characteristics of the topological compliance parameters, and generating a spatial security constraint model in combination with local spatial medium attributes; and S5, based on the spatial security constraint model, outputting structure abnormal information and corresponding optimized path planning parameters. And through multi-channel tensor quantization variation characteristics, safety constraints are constructed by using an anisotropic bounding box, and a path is optimized in combination with a potential field algorithm, so that high-precision automatic obstacle avoidance is realized.

According to the scheduling method for reducing the receiving jitter of the transport streams on the disjoint redundant routing in the time-sensitive network, a heuristic method is provided based on a TSN circular queue forwarding shaper and an IEEE 802.1 CB protocol, and the purpose is to minimize the average jitter of all data streams and reduce the receiving jitter of the transport streams on the disjoint redundant routing. The time slot with the low utilization rate on each redundant route is selected through calculation, so that network congestion is relieved, and receiving jitter is reduced. And a time slot reducing program is adopted in the heuristic algorithm, and after all flow scheduling is finished, the size of the time slot is reduced by calculating residual unused resources in each time slot of all links in the topology in a super cycle, so that the receiving jitter is further reduced, and on the premise that the schedulability is kept, the scheduling efficiency is improved. And the reliability is ensured and the receiving jitter is reduced.

The invention relates to the technical field of bridge construction safety monitoring, and discloses a road and bridge construction multi-sensor data fusion safety early warning system. The system comprises a data preprocessing and quantum state characterization module, a heterogeneous tensor field construction and quantum correlation analysis module, a quantum topological structure analysis and risk area identification module, a non-equilibrium tensor field evolution and risk early warning module and an early warning output and visualization module. The method comprises the following steps: uniformly mapping multi-source heterogeneous sensor data into a density matrix in a high-dimensional Hilbert space, calculating quantum correlation degree quantity among the data to construct a multi-dimensional heterogeneous tensor field and a quantum correlation topological graph, and identifying a risk area and a propagation path by calculating topological invariants of the topological graph. And the risk critical point is predicted by monitoring entropy generation rate evolution of the tensor field. According to the method, the heterogeneous data can be deeply fused, the weak links of the structure can be identified from the system level, the risk evolution trend can be prospectively pre-judged, and the early-stage and interpretable early warning of the construction risk can be realized.

The invention discloses a power system feature extraction method and device, electronic equipment and a storage medium, and is used for solving the problems that a dimension reduction result cannot reflect an actual physical relationship due to physical information loss, parameters before and after dimension reduction are lack of clear association, and the adaptability limitation is large in a traditional feature extraction method. The method comprises the following steps: acquiring power operation data of a power system, and extracting original high-dimensional features of the power operation data; the original high-dimensional feature corresponds to a plurality of samples, and each sample corresponds to one supervision signal; according to the supervision signal of each sample, physical constraint dimension reduction optimization is combined, original high-dimensional features are mapped to a low-dimensional feature space in a dimension reduction mode, and dimension reduction optimization features are obtained; calculating a topology retention rate based on the original high-dimensional features and the dimensionality reduction optimization features; and when the topology retention rate meets a physical retention condition, outputting a dimension reduction optimization feature for performing stability analysis on the power system.