2420 results about "Abnormality detection" patented technology

The invention discloses a distributed slope monitoring system based on edge cloud collaborative intelligent adaptive decision. The distributed slope monitoring system comprises a plurality of intelligent sensing unit ISU nodes deployed at key positions of a slope and a data processing and intelligent analysis unit, each intelligent sensing unit ISU node is used for transmitting data to an edge gateway in an ad hoc network wireless mode or directly uploading the data to a cloud platform and carrying out slope monitoring based on a local adaptive monitoring strategy; the data processing and intelligent analysis unit comprises an edge intelligent module, a cloud gateway and an edge gateway; the edge gateway serves as a middle layer and is used for protocol conversion, data aggregation, temporary storage and preliminary analysis; the cloud gateway is used for providing calculation and storage resources, training a more complex AI model based on historical and real-time data, and performing pattern recognition, prediction analysis and anomaly detection tasks; the edge intelligent module comprises a plurality of edge computing nodes and is internally provided with a lightweight AI reasoning unit, and the edge intelligent module is arranged on the edge side and used for implementing edge intelligent processing.

The invention relates to the technical field of biomedical engineering, in particular to a multi-mode brain anomaly detection method and system based on machine learning. The method comprises the following steps: acquiring brain medical image data of different modalities, and realizing spatial registration and alignment through a multi-modal registration algorithm based on mutual information; a multi-branch feature extraction model including a convolutional neural network, a converter and a state space model is utilized to perform feature embedding on the original image of each modal; performing frequency decoupling on the features of each mode through adaptive approximate wavelet transform, and decomposing the features into high-frequency detail information and low-frequency global information; a frequency band fusion strategy based on an attention mechanism is implemented on high and low frequency features of different modal images, and fused frequency sub-band features are input into a space-frequency Mama module. Through the adaptive frequency domain decomposition and cross-modal fusion mechanism, the multi-modal brain image information is effectively integrated, and the accuracy and robustness of brain anomaly detection are remarkably improved.

The invention discloses a power distribution room anomaly detection system based on cloud side-end cooperation, and belongs to the technical field of intelligent power grids. In order to solve the problems of high network bandwidth pressure, insufficient edge computing capability, low anomaly detection accuracy, difficulty in multi-source data fusion and the like caused by the adoption of an end-cloud direct connection architecture in an existing power distribution room monitoring system, the system comprises: a data acquisition layer configured with various heterogeneous sensors to acquire operating parameters and environmental data in real time; the edge storage and calculation layer carries out local real-time processing, anomaly detection, model training and visual display, an anomaly detection module of the edge storage and calculation layer carries out research and judgment on real-time data to generate early warning information, and a prediction and detection linkage module monitors an anomaly probability trend and adjusts a sampling frequency; the edge gateway realizes protocol conversion and data forwarding; and the cloud decision-making layer aggregates multi-edge node data, optimizes a global model through federal learning, and issues and updates a local model. The system is used for improving the accuracy, real-time performance and reliability of anomaly detection of the power distribution room, reducing the operation and maintenance cost and realizing intelligent operation and maintenance.

The invention relates to the technical field of quality detection, in particular to a product quality control method and system based on machine vision, and the method comprises the following steps: obtaining product surface image data, calculating the gray gradient value of each pixel, extracting the gray gradient change rate, recording the gradient amplitude and direction information, and generating product surface gradient data. According to the method, through pixel-level gray scale gradient calculation, the product local feature expression ability is improved, multi-scale gradient change trend analysis is combined, the accurate recognition ability of a product defect area is improved, through texture direction angle calculation and vector field construction, the direction change anomaly detection reliability is enhanced, and the direction change anomaly detection accuracy is improved based on the combination of a direction deviation accumulated value and an abrupt change threshold value. Effective identification of a structure sudden change area is ensured, adjustment is carried out for curvature continuity abnormal points, defect boundary fitting precision is optimized, defect area internal gradient distribution and boundary feature comparative analysis are carried out, accurate classification of defect types is realized, and stability and adaptability of automatic product quality detection are ensured.

The invention discloses an unmanned aerial vehicle cruising method and system based on a deep learning artificial intelligence image recognition algorithm. According to the method, an unmanned aerial vehicle carrying an improved YOLOv7-SwinT target recognition model collects real-time image data of an inspection area, and the model fuses a single-stage target detection architecture of YOLOv7 and a visual feature extraction network of Swin Transform. Progressive target detection is realized by adopting a three-level recognition architecture, wherein the progressive target detection comprises primary anomaly detection based on lightweight CNN, intermediate accurate positioning in combination with an attention mechanism and advanced target classification of multi-sensor data fusion. And the system combines the electric quantity of the unmanned aerial vehicle, the environmental condition and the task priority according to the identification result, generates a dynamic inspection path through an adaptive path planning algorithm, and realizes multi-vehicle collaborative operation by using an intelligent task allocation algorithm. In the inspection process, sensor data are processed in real time through edge computing equipment, and charging scheduling is optimized by adopting an intelligent energy management system. The target recognition precision and the cruising efficiency of the unmanned aerial vehicle in a complex environment are remarkably improved, and the method is suitable for application scenes such as electric power inspection and security monitoring.

The invention relates to the technical field of electric energy meter metering, and discloses an electric energy meter metering anomaly analysis method and system, and the method comprises the steps: collecting data, such as voltage waveforms, to generate a metering feature vector set, and constructing an anomaly detection rule base; setting a scene parameter type set, and establishing an abnormal association judgment model; dynamically correcting the threshold value by combining the model, and generating an optimized error threshold value set and an abnormal triggering condition set; and updating a metering analysis strategy to generate an abnormal judgment scheme, and correcting data verification sequential logic. The system comprises a data acquisition module, an abnormal rule base construction module, a scene parameter configuration module, a correlation model training module, a dynamic threshold optimization module, a strategy updating module and a time sequence correction module. Through multi-dimensional data modeling, scene-based threshold configuration and dynamic time sequence calibration, the accuracy and adaptability of electric energy meter measurement anomaly detection are improved, and the method is suitable for measurement anomaly analysis of diversified power consumption scenes in a smart power grid.

The invention discloses a security monitoring video intelligent analysis method based on multi-algorithm cooperation and unified architecture, and relates to the technical field of intelligent video monitoring, and the method comprises the steps: collecting a security monitoring video, carrying out the preprocessing, carrying out the spatial-temporal feature extraction and fusion through a CNN-LSTM spatial-temporal fusion engine, and outputting a fusion feature map; performing target detection, behavior recognition and anomaly detection on the fused feature map to obtain a multi-algorithm analysis result; a cross-module fusion mechanism based on an attention mechanism is utilized to perform weighted integration processing on the multi-algorithm analysis result to obtain a fusion event representation vector; performing event type identification and risk level evaluation on the fusion event representation vector to obtain an event classification result and an event risk level; according to the invention, through the CNN-LSTM space-time fusion engine, the front-end perception capability of abnormal behaviors in a complex scene is improved, and the event detection accuracy and the anti-interference capability are improved.

The invention relates to the technical field of user power utilization chain analysis, and discloses a special transformer user power utilization abnormal chain construction method fusing deep learning, which comprises the following steps: collecting time sequence power utilization data and power grid topological data of special transformer users, taking each special transformer user as a node, constructing a graph structure comprising a physical connection edge and a behavior association edge, and constructing a graph structure comprising a physical connection edge and a behavior association edge; and inputting the constructed graph structure into a space-time graph neural network model, introducing a power grid physical constraint condition into an optimization target, taking a power grid operation rule as a constraint embedding model, and outputting node-level, edge-level and sub-graph-level multi-level anomaly detection results. Constructing a heterogeneous graph containing user nodes, anomaly type nodes and time slice nodes through the multi-level anomaly detection result, performing path search in the heterogeneous graph through a predefined association mode template, generating a candidate anomaly chain, performing causal strength verification on the candidate anomaly chain by adopting a time sequence causal relationship verification model, and obtaining a multi-level anomaly detection result of the user nodes, the anomaly type nodes and the time slice nodes. And intelligent analysis of the user electricity consumption abnormity chain is realized.

The invention discloses a data ownership verification and authorization control method based on zero knowledge proof. The method comprises the following steps: constructing a traceable Merk tree structure on a block chain or a distributed account book; a data owner calculates data fingerprints and packages the data fingerprints into declaration nodes to be inserted into the traceable Merk tree; storing the ownership zero knowledge proof and the traceable Merk tree root hash into a block chain together; the verification party executes the zero-knowledge verifier contract on the chain to complete ownership verification without reading original data or identity plaintext; access control is realized after on-chain verification of the service party; triggering local heavy hash to update the traceable Merk tree root hash, and broadcasting a revocation event on a chain; and confirming legality of evolution and authorization change of all nodes. According to the method, the historical traceability and structural consistency of the data ownership and authorization relationship are ensured, and efficient traceability and anomaly detection of the node evolution chain are also realized.

The invention discloses a detection system for operation abnormity of an electric drive axle motor of a new energy commercial vehicle, and belongs to the technical field of motor abnormity detection. The system comprises a multi-source sensing module, a dynamic coupling analysis module, a harmonic distortion traceability module, a dynamic stability analysis module, an anomaly fusion decision module, a self-adaptive threshold generation module and a fault tree reasoning module. Multi-dimensional signals such as current, vibration, temperature and a rotor position angle are synchronously acquired through the multi-source sensing module, and multi-physics field collaborative analysis is performed through the dynamic coupling analysis module and the harmonic distortion traceability module, so that the fault diagnosis accuracy of operation abnormity of the electric drive axle motor of the new energy commercial vehicle is improved; the problem that in the prior art, detection only depends on a single physical quantity, and misjudgment or missing detection is easily caused by interference is solved.

The invention discloses a fruit and vegetable juice production line equipment fault tracing method, which comprises the following steps of: acquiring parameters such as temperature, pressure, vibration, rotating speed and motor current in real time through a multi-channel sensor, and establishing a working condition characteristic database by combining filtering, normalization, statistics and frequency domain characteristic extraction; based on a support vector regression algorithm, a nonlinear mapping model of working condition features and anomaly detection thresholds is constructed, and dynamic threshold adaptive output and real-time anomaly judgment for different working conditions are realized; according to a detection result, a fault signal is automatically triggered, a model is continuously incremented and trained, the adaptability to new working conditions is improved, the accuracy, intelligence and stability of equipment anomaly detection are effectively improved, misinformation and missing information can be reduced, and the automatic operation and maintenance level of a production line is enhanced.

The invention relates to the technical field of machine learning, and provides a building equipment abnormity identification method and system based on machine learning, which are used for realizing accurate detection and accurate early warning of building equipment abnormity. The method comprises the following steps: acquiring a continuous operation data set of target building equipment, wherein the continuous operation data set comprises multiple segments of equipment state recording units with timestamp marks; performing time-frequency domain feature extraction processing on the continuous operation data set to obtain a time-frequency domain feature set of the equipment state recording unit; calling a pre-constructed hybrid machine learning model to perform anomaly detection processing on the time-frequency domain feature set, and generating an anomaly recognition result of the equipment state recording unit; determining the anomaly type of the target building equipment and the distribution feature information of the anomaly type in the time dimension according to the anomaly recognition result; and generating a target early warning instruction containing a time positioning identifier based on the exception type and the time distribution feature information, and sending the equipment early warning instruction to a target equipment management terminal.

The invention discloses a new energy equipment intelligent operation and maintenance system and method based on digital twinning, and relates to the technical field of new energy equipment operation and maintenance management. The system comprises a data acquisition module, a digital twin construction module, a model adaptive module, an intelligent analysis module, a decision optimization module and a knowledge closed-loop module. The data acquisition module realizes multi-source heterogeneous data fusion and standardization; the digital twin construction module generates geometric, physical and behavior models and is linked with real-time data; the model adaptive module dynamically calibrates the key parameters; the intelligent analysis module completes anomaly detection, fault diagnosis and life prediction; the decision optimization module formulates a maintenance scheduling and operation strategy and realizes a control closed loop; and the knowledge closed-loop module constructs a structured fault graph through text analysis to continuously optimize the model. The method can be widely applied to wind power, photovoltaic, energy storage and other scenes, and efficient and intelligent operation and maintenance of new energy equipment are achieved.

The invention discloses an intelligent automobile interpretable abnormity diagnosis method and system, and relates to the technical field of intelligent traffic. The method comprises the steps of collecting multi-dimensional sensor data based on an intelligent automobile test platform, and constructing a directed causal graph and a causal adjacency matrix which are used for describing a causal relationship between the sensor data; designing a causal constrained graph attention mechanism based on the causal adjacency matrix, and constructing a causal constraint enhanced graph attention anomaly diagnosis model; and based on the directed causal graph and the graph attention anomaly diagnosis model, constructing a hierarchical anomaly diagnosis strategy integrating a feature reconstruction error, a variable causal relationship and a graph attention network weight, positioning an anomaly root cause and identifying a propagation path of the anomaly in the sensor network. According to the invention, the problems of false correlation and lack of exception explanation ability of graph attention network learning in the prior art can be overcome, and reliable exception detection and root cause diagnosis of intelligent automobile multi-sensor data are realized.

The invention provides a tumor electric field treatment system, an electrode slice temperature detection method and an electrode slice anomaly detection method, the system comprises an alternating current signal generator, an electrode slice and a controller, the electrode slice comprises a plurality of electrode units and a plurality of temperature detection units, the plurality of temperature detection units are divided into a plurality of row groups and a plurality of column groups, and the plurality of electrode units are connected with the controller. The grounding ends of the temperature detection units in each row group are jointly connected to the same grounding wire, and the signal ends of the temperature detection units in each column group are connected to the same dual-purpose signal wire after being in short circuit with the corresponding electrode units; when the dual-purpose signal line is connected to the temperature sampling point, the grounding wires are sequentially conducted, so that temperature detection signals detected by the temperature detection units are sampled based on the temperature sampling point; when the dual-purpose signal line is connected to the alternating power line, alternating electric signals are transmitted to the electrode units. Therefore, the temperature of the electrode plate can be detected by using fewer conductive traces, and whether the electrode plate is abnormal or not can be judged through temperature detection.

The invention provides an anomaly detection method and device for image data and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: carrying out the feature extraction of the image data, fusing the spectrum and spatial features to obtain a joint feature matrix, and inputting an anomaly detection model; the model adopts an alternating direction multiplier algorithm to solve a low-rank sparse decomposition problem, and an objective function comprises a data fidelity item, a regularization item and a waveband weight item; the regularization item comprises a low-rank constraint and a sparse constraint, and the wave band weight item acts on the low-rank constraint in a weighting form; updating a background low-rank tensor, an abnormal sparse tensor, a Lagrange multiplier, a sparse constraint weight, a wave band weight item and penalty parameters of an algorithm by adopting an iteration mode in a solving process; and repeating iteration until a preset termination condition is reached, calculating an abnormal score graph pixel by pixel based on the abnormal sparse tensor, and comparing to determine an abnormal target. The problem that an abnormal target is difficult to accurately recognize in a complex scene can be solved, and detection precision and efficiency are improved.

The invention relates to the technical field of mutual inductor tests, and provides a mutual inductor test abnormal data automatic filtering analysis method and system, and the method comprises the steps: obtaining magnetic field data, electric field data and thermal field data, respectively carrying out the feature extraction and weighted fusion, and obtaining a multi-physical field fusion feature vector; constructing an electromagnetic induction chain type propagation graph, calculating the weight of an edge in the graph, constructing a magnetic flux conservation constraint graph convolutional neural network, learning propagation and evolution modes of transformer test abnormity, and obtaining a coupling abnormity feature vector; the coupling anomaly feature vectors are classified, normal data, single physical field abnormal data and multi-physical field coupling abnormal data are recognized, and corresponding data are judged as abnormal data and filtered; and carrying out physical mechanism analysis on the filtered abnormal data to obtain an abnormal detection report. According to the invention, high-precision automatic identification, filtering analysis and physical mechanism traceability diagnosis of transformer test abnormal data are realized.

The invention belongs to the technical field of integrated circuits, and particularly relates to a speed-adjustable fan driving and anomaly detection circuit with high reliability and low cost. Comprising a current expansion driving module for providing a voltage FCC stabilized by an LDO for a fan and performing current expansion on an output current of the LDO to drive the fan; the logic control module is connected with the current expansion driving module and is used for adjusting a control signal ADJlt; n: 0gt; the value of the voltage FCC output by the LDO is changed; the open circuit and locked-rotor detection module is connected with the logic control module and the VIN port of the LDO; and the short circuit detection module is connected with the logic control module and the output end of the LDO, outputs a signal to a state flag bit FTO, and judges whether the fan is in a normal working state, an open circuit state, a locked-rotor state or a short circuit state or not according to the waveform and the level of the state flag bit FTO. The speed-adjustable fan driving circuit is integrated, and the function of detecting various abnormal states of the fan is achieved.

The invention discloses a battery change cabinet battery abnormity detection method and system based on a thermal image, and relates to the related field of battery safety monitoring, and the method comprises the steps: collecting a surface thermal image sequence in a battery compartment through an infrared thermal imager arranged in a battery change cabinet, and enabling the surface thermal image sequence to have a time sequence and a charging stage label; extracting temperature field features of a space-time dimension from the thermal image sequence, and identifying an abnormal state and an abnormal region of the battery based on the temperature field features; and acquiring real-time battery management system (BMS) data of the battery, performing correlation verification on the abnormal state or the abnormal area and the BMS data, and determining an abnormal level and positioning a fault battery unit according to a correlation verification result. According to the invention, the technical problems of inaccurate and untimely detection and inaccurate fault positioning in the existing battery abnormality detection of the battery changing cabinet are solved, and the technical effects of improving the accuracy, timeliness and positioning accuracy of battery abnormality detection are achieved.

The invention discloses a method for detecting abnormal metering performance of an intelligent electric energy meter, which belongs to the technical field of electric energy metering equipment and comprises the following steps of: 1, establishing a reference response curved surface of parasitic parameters of a voltage sampling resistor and a current transformer in the electric energy meter relative to temperature and humidity; step 2, during the operation period of the electric energy meter, acquiring the real-time measurement value of the parasitic parameter in situ; and step 3, matching the real-time measurement value with the reference response curved surface. According to the method, an active matching model library can be constructed by fusing an underlying physical model of a component and a mapping rule set for a complex environment with sudden change of plateau outdoor temperature and humidity, the internal mechanism that parasitic parameters change along with temperature and humidity is explained from the physical essence level, and the limitation of a pure data statistical method is made up; the dynamic error judgment threshold value can adapt to environmental stress changes in real time, and the accuracy of metering performance anomaly detection and the complex environment adaptability are effectively improved.

The invention discloses a power distribution cabinet cable monitoring method, system and device based on edge calculation, and relates to the technical field of power system monitoring and early warning, and the method comprises the steps: arranging a sensor in a power distribution cabinet, and collecting the displacement and deformation data of a cable; carrying out preprocessing and preliminary analysis on the collected data by utilizing edge computing equipment; the method comprises the following steps: determining whether cable displacement deformation is in an abnormal state or not according to a preset judgment condition through edge computing equipment, and uploading abnormal data to a cloud server if the cable displacement deformation is judged to be abnormal; and the cloud server receives the abnormal data and carries out deep analysis in combination with historical data. Through rapid processing of the edge computing device, once the cable is abnormal, the system can timely perceive and upload abnormal data, the real-time performance of monitoring is remarkably improved, it is ensured that the state change of the cable can be rapidly captured, meanwhile, a deep learning algorithm is introduced, more accurate anomaly detection is achieved, and the accuracy and reliability of early warning are improved.

The invention discloses a column type distribution transformer winding abnormity detection system and detection method. The detection system comprises a frequency response detection instrument, a signal coupling assembly, an impedance matching module and a data processing module. The frequency response detection instrument is used for outputting a frequency sweep excitation signal and collecting a winding response signal; the signal coupling assembly is used for isolating power frequency voltage and filtering low-frequency interference; the impedance matching module is used for adjusting output impedance and input impedance to realize matching with electrical characteristics of different windings; and the data processing module is used for constructing an equivalent detection loop model, completing frequency response simulation calculation and carrying out frequency domain analysis and anomaly identification on an actually measured response curve. According to the invention, electrified state diagnosis of distribution transformer winding structure abnormity (such as deformation and loosening) can be realized.

The invention provides a medicine bottle array detection system based on vision and sensing fusion, and relates to the technical field of data processing, and the system comprises the steps: obtaining the spatial distribution data of a vision unit and the posture state data of a sensing unit, fusing the position parameters and state parameters of each bottle body, building a multi-dimensional data node under a unified time index, calculating the spatial aggregation degree of the medicine bottle array to obtain an array quantity, combining the aggregation change of each time point into an evolution chain table, carrying out joint operation on the difference result and the state parameter of adjacent nodes in the evolution chain table, calculating the dynamic coupling degree of the medicine bottle array to obtain a sequence potential energy value, and dividing the sequence potential energy values of the different numerical value intervals into a plurality of level nodes, and performing consistency analysis and anomaly detection on features of the different level nodes to obtain detection result data. By detecting the state of the medicine bottles and the arraying direction, bottle toppling early warning is achieved.

The invention discloses an urban construction project whole-process quality safety closed-loop supervision and tracing platform and method, and relates to the field of data tracing, and the method comprises the steps: constructing a project whole-life-cycle digital twin based on urban construction project data, and presetting a quality safety standard parameter set; implanting an Internet of Things traceable unit in the urban construction project key entity, and mapping the association relationship and the entity to a digital twinborn body; processing data acquired by the sensor network to generate a structured engineering state data set; an anomaly detection model is constructed based on CNN and LSTM, and when data deviates from a threshold value, early warning is triggered and a responsibility chain is traced; generating a processing instruction tree, and distributing the processing instruction tree to a responsibility subject; and the responsibility subject uploads a disposal process and a result, binds the disposal process and the result with an early warning event ID after Hash encryption, and stores the disposal process and the result in a traceability database. The method has the advantages that efficient quality safety supervision and tracing of the whole urban construction project process are achieved through the digital twinborn bodies, and the project safety and management efficiency are improved.

The invention discloses a complex metering sensor abnormal feature analysis method based on long-term accumulated data, and relates to the technical field of state monitoring and fault prediction. Static statistical characteristics and dynamic frequency characteristics of data and transient changes in non-stationary signals can be comprehensively captured, meanwhile, a model in which a double-layer long-short-term memory network is combined with an attention mechanism is constructed, short-term time sequence dependence in a first-layer LSTM learning data fragment and long-term evolution trend between second-layer LSTM learning fragments are constructed, and the time sequence dependence in a second-layer LSTM learning data fragment is constructed. The attention mechanism focuses on the key period, and the deep fusion from feature extraction to model construction enables the model to accurately identify the difference between normal data and abnormal data, thereby realizing the accurate detection of the sensor abnormality, and in the practical application, the normal fluctuation and real abnormality of the sensor can be effectively distinguished, and the accuracy of the sensor abnormality detection is improved. And reliable guarantee is provided for stable operation of the system.

The invention discloses a real-time data analysis method and system based on multi-modal semantic mapping, and the method comprises the steps: obtaining input event information containing a multi-modal data flow, and carrying out the preprocessing of the multi-modal data flow; performing space-time alignment on each piece of modal data in the preprocessed multi-modal data stream, extracting modal features of each piece of modal data by using a pre-trained multi-modal encoder, and mapping the modal features to a unified semantic space; dynamically adjusting the weight of each modal feature by using a routing network according to the event information, and fusing each modal feature based on the weight to obtain a fused feature; and dynamically analyzing the cluster structure change of the fusion feature distribution by using a clustering algorithm, carrying out anomaly detection by using an outlier analysis method according to the cluster structure change, and if an anomaly triggering condition is met, generating an alarm signal.

The invention relates to the technical field of data processing, and provides a communication bus anomaly detection method and device, electronic equipment and a readable storage medium. The method comprises the steps of collecting multi-dimensional communication parameters of at least one communication bus in a vehicle, wherein the multi-dimensional communication parameters comprise protocol layer parameters and physical layer electrical parameters; the multi-dimensional communication parameters are input into a health degree scoring model, the health degree score of the communication bus is obtained, and the health degree scoring model is obtained through training according to historical communication parameter data; determining a state level of the communication bus according to a comparison result of the health degree score and a preset threshold value; when the state level is abnormal, performing matching with a predefined abnormal mode feature library based on the multi-dimensional communication parameters to obtain an initial abnormal type; and performing consistency verification according to the initial exception type and the state verification data from different data sources to obtain a target exception type of the communication bus.

The invention discloses a power switch abnormity monitoring method and system based on artificial intelligence, and relates to the technical field of power communication network monitoring, and the method comprises the steps: collecting multi-source operation data of a switch and a port in a power communication network, and constructing a high-dimensional feature tensor through time synchronization, data alignment, missing compensation and residual extraction; outputting a node residual error based on a self-supervised prediction model, extracting a topological consistency feature in combination with a dynamic threshold and a graph attention mechanism, and constructing a multi-source feature through a comprehensive abnormal sub-model in combination with residual error intensity, a prototype distance and an extreme value tail risk; and setting a grading alarm threshold value based on dynamic distribution, and outputting an alarm optimization strategy through abnormal aggregation and topological consistency analysis. According to the method, through multi-source feature modeling, self-supervision prediction and comprehensive anomaly division, anomaly accurate identification, topology consistency analysis and dynamic alarm optimization are realized, the multi-domain data fusion capability and anomaly detection precision are improved, and the method has innovativeness and engineering application value.

The invention discloses a method, a system and equipment for predicting the stability of integrated circuit test equipment and a medium, and belongs to the technical field of integrated circuit test. The method comprises the following steps: firstly, carrying out preprocessing and feature selection on historical data in an FT test stage, and adopting a Gaussian mixture model (GMM) to cluster and identify different operation condition clusters of a test machine; then, establishing a health state GMM reference for each working condition cluster, calculating a KL divergence value of a normal sample and the reference, and setting a dynamic anomaly detection threshold by 99.73% quantile of the KL divergence value; and finally, in real-time monitoring, calculating a KL divergence value of real-time data and a corresponding working condition cluster benchmark, and comparing the KL divergence value with a dynamic threshold value to realize accurate anomaly marking. The method effectively solves the problem of abnormal detection of the test data of the integrated circuit under complex and changeable working conditions, and improves the monitoring accuracy and working condition adaptability.

The invention discloses a manufacturing task autonomous negotiation and execution method based on a large language model agent, and the method comprises the steps: constructing a production scheduling agent, an equipment management agent, a material distribution agent and a quality control agent, analyzing a natural language task instruction through the production scheduling agent, and decomposing the natural language task instruction into subtasks; each agent calculates a utility value based on the load rate, the resource matching degree, the estimated completion time and the historical success rate, and performs structured negotiation to achieve a task allocation consensus; a prediction-check-rollback architecture is adopted to generate an action instruction sequence, the sequence is compiled into a time Petri network transition sequence, and reachability verification is carried out based on hard security constraints; production environment data is collected in real time to trigger anomaly detection and re-negotiation, and a formalized security verification and causal anti-factual reasoning parameter updating mechanism is introduced. According to the method, unstructured instruction understanding, autonomous task planning, multi-agent collaborative decision and closed-loop optimization are realized, and the problems of real-time performance, safety and interpretability of a large language model in manufacturing control are solved.