189 results about "Decision networks" patented technology

The invention provides a large model lightweight reasoning deployment method under limited hardware resources, and the method comprises the steps: quantifying the weight importance of a large model through a composite index of gradient sensitivity and activation frequency, and carrying out pruning operation in combination with an improved index weighted moving average strategy, thereby obtaining a structured sparse model; the sparse model is divided into sub-networks by adopting double rules, a routing decision network is trained, and an adaptive feature shunting architecture model is constructed; a multi-precision weight set is generated through a nested quantization technology, quantization bit width is dynamically adjusted, and edge equipment hardware parameters are adapted to complete reasoning environment initialization; after a reasoning request is received, an optimal sub-network is selected based on the trained routing decision network, corresponding weights are loaded in parallel, and a reasoning result is fused and output; and converting a reasoning result format, and dynamically optimizing a scheduling strategy based on a system real-time monitoring index. The method is compatible with a mainstream large model and a hardware platform, and an efficient and universal deployment scheme is provided for end-side AI engineering landing.

The invention discloses a multi-energy-storage thermal power generating unit load prediction and optimal scheduling method and system, and relates to the technical field of multi-energy-storage thermal power generating units, and the method comprises the steps: collecting the operation parameters and external environment parameters of a thermal power generating unit in real time, and constructing a real-time state parameter matrix; constructing a load prediction model based on the historical state parameter matrix, importing the real-time state parameter matrix into the load prediction model, outputting a load trend prediction curve, and triggering an early warning signal through a secondary discrimination mechanism; identifying a load disturbance value based on the load trend prediction curve, obtaining a load disturbance sequence, and decoupling the load disturbance sequence into a plurality of components; inputting the plurality of vectors into a preset decision network, dynamically correcting a constraint condition built in the decision network in combination with the early warning signal, introducing an improved dragonfly algorithm for optimization iteration, and generating an optimization scheduling instruction; according to the method, the adaptability of optimal scheduling and high-precision prediction of the load trend are improved.

The invention discloses a knowledge graph-based old building reconstruction scheme automatic recommendation method. The method comprises the following steps of S1, obtaining and preprocessing old building house data; s2, extracting a semantic entity and attribute relationship from a transformation case library, a building specification library and a construction scheme library, and constructing a knowledge graph; s3, mapping the house data to knowledge graph entity nodes, and executing graph embedding to generate building semantic representation; s4, constructing a graph neural network, calculating node semantic relevancy, and generating a building state vector; s5, inputting the building state vector into the reinforcement learning decision network, and optimizing the strategy to obtain an optimal transformation action; s6, screening reconstruction measures from the knowledge graph according to the optimal reconstruction action, and performing scoring to form candidate schemes; and S7, sorting the candidate schemes, selecting the scheme with the highest score, and recommending and updating the knowledge graph. According to the method, intelligent generation and self-optimization recommendation of the old building reconstruction scheme are realized, and the reconstruction efficiency of the scheme is remarkably improved.

The invention relates to the technical field of intelligent switching, in particular to a multi-mode dynamic coupling intelligent switching method and system based on road condition recognition and a vehicle. The method comprises the steps that gradient, vehicle speed, accelerator and brake signals are collected in real time, and a multi-dimensional feature vector fusing driving habits is constructed after preprocessing; a real-time road condition label is automatically generated through unsupervised clustering, and the clustering effectiveness is verified online by using a contour coefficient. Enabling the road condition label and the current power mode to form a system state, and inputting the system state into a self-learning decision network based on reinforcement learning; and the network updates the state-action value matrix through a reward function in a time sequence difference mode, and outputs a target power mode with the highest accumulated reward. And calculating the motion confidence coefficient of the target mode based on the value matrix, if the motion confidence coefficient exceeds a threshold value, generating a switching instruction, and smoothly adjusting the torque by an execution mechanism through a torque transition algorithm to realize impact-free switching. Automatic road condition recognition, self-adaptive decision making and non-inductive execution are achieved, and smoothness and adaptability are improved.

A method and a system of bearing fault diagnosis based on a dual attention mechanism to strengthen a hierarchical decision network are provided, where the method includes the following steps: collecting bearing vibration signals in different health states; based on the bearing vibration signals, constructing a hierarchical multi-class fault diagnosis model; and determining a fault position and a fault size of a bearing by using the hierarchical multi-class fault diagnosis model.

The invention discloses a bearing fault diagnosis method based on time-frequency double-flow complementation and adaptive gating fusion, and the method comprises the steps: collecting an original vibration signal of a bearing, segmenting an original vibration signal sequence into a plurality of sample segments with fixed lengths, and forming an original data set; dividing the original data set into a training set and a test set by adopting a physical segmentation strategy based on a time axis; constructing a time-frequency double-flow feature extraction module, and extracting a time-domain feature vector and a frequency-domain feature vector; constructing an adaptive gating fusion module to dynamically generate a gating weight, and performing weighted complementary fusion on the time domain feature vector and the frequency domain feature vector to obtain a fused fault feature vector; constructing a classification decision network to map the fusion fault feature vector to a decision space, and optimizing model parameters by using a label smooth regularization loss function and a cosine annealing strategy; and inputting the test set into the trained fault diagnosis model, calculating posterior probability distribution of a sample belonging to each fault category, and outputting a bearing fault diagnosis result.

The invention relates to the technical field of equipment fault prediction and diagnosis, and discloses an equipment fault prediction and diagnosis method and system based on dynamic feature fusion. Comprising the steps of collecting a multi-modal monitoring signal, preprocessing to generate a standardized time series data stream, extracting a feature vector through a local detail sensing network and a global trend capturing network, generating a comprehensive feature vector by using an adaptive feature fusion module, and outputting an operation state category and a confidence score through a classification decision network. According to the method, the multi-modal time sequence features can be effectively fused, the feature weight is dynamically adjusted, the equipment operation state is accurately identified, and the accuracy and reliability of fault prediction and diagnosis are improved.

The invention discloses a cross-domain data collaborative management method based on edge computing. The method comprises the following steps: S1, acquiring data to form a time series data set; s2, extracting features of time sequence data at edge nodes, and generating a state vector sequence; s3, inputting the state vector sequence into an edge decision network for weighted aggregation to generate a collaborative state sequence; s4, detecting a behavior change point of the collaborative state sequence by adopting a cumulative sum control chart method, and constructing a state mapping sequence; s5, based on the state mapping sequence, constructing a region index set through discrete wavelet transform; s6, inputting the regional index set into an edge decision network to generate a management instruction sequence; and S7, executing the management instruction sequence at the edge node, and updating the edge decision network based on an execution result. According to the method, multiple intelligent algorithms are fused to realize data collaborative management of cross-domain edge nodes, and the response efficiency and decision-making ability in an edge environment are improved.

The invention discloses an aircraft complex terrain dynamic path planning method and system based on reinforcement learning, and the method comprises the steps: S1, carrying out the multi-source heterogeneous perception fusion, and generating a state vector through the multi-source perception fusion; s2, element reinforcement learning intelligent decision making: dynamically adjusting the weight of a reward function by utilizing an element reinforcement learning agent and outputting an action instruction; s3, a safety control instruction is generated, model prediction control is adopted to conduct rolling optimization on the instruction under the condition that dynamics and safety constraints are met, and a final control instruction is generated; and S4, online learning and strategy updating: storing interaction experience through an online learning mechanism, preferentially sampling high-complexity scene data, and finely adjusting decision network parameters. According to the method, the dynamic adaptive path planning of the aircraft in the complex terrain is realized, the defects of poor adaptability and dependence on prior data in the prior art are overcome, and the autonomy and safety of path planning are improved.

The invention belongs to the technical field of power grid dispatching, and discloses an energy system low-carbon dispatching method and device based on deep reinforcement learning, and the method comprises the steps: changing the input structure of a decision network through the extraction of Riemannian flow pattern features and the analysis of group distribution; the input of a global decision network in the prior art is changed from environment state information to global state information and streaming feature group distribution and prediction group distribution at the next moment. The input of a single local decision network in the prior art is changed from local state information to flow pattern feature group distribution, predicted group distribution at the next moment and a corresponding unit state; according to the method, the input dimension of the decision network and the complexity of decision search are remarkably reduced, so that the algorithm efficiency is effectively improved, and rapid scheduling of the energy system is realized.

The invention belongs to the technical field of communication, and particularly relates to a router system service exception self-healing method based on cloud AI, which comprises the following steps: deploying an exception detection module at a router end, collecting system logs and state data in real time and generating a standardized exception report; uploading the report to a cloud AI server through an encrypted MQTT protocol; the cloud calls a hybrid analysis model composed of a rule matching engine, a machine learning classifier and a reinforcement learning decision network to generate a self-healing strategy instruction packet; the router end receives and executes the strategy, completes service restart, configuration rollback or hotfix loading and other operations, and verifies the self-healing effect; when communication interruption exceeds a threshold value, an embedded loopback self-healing subsystem is automatically activated, and abnormity is independently handled based on a local strategy library. According to the technical scheme, millisecond-level abnormal response and high-success-rate autonomous recovery can be achieved, the network availability and the service continuity are remarkably improved, and the disaster recovery self-healing capacity is still achieved when the cloud end is disconnected.

The invention relates to the technical field of agricultural engineering, and particularly discloses a rice bud root length intelligent classification system and method based on DINOV3 and SAM algorithms. In order to solve the problems that traditional rice bud length detection depends on manpower, is low in efficiency and high in subjectivity, and an existing automatic detection method is limited in feature extraction capacity, poor in sample imbalance adaptability and only capable of achieving single task classification, the method fuses a self-supervised visual model DINOv3 and a universal segmentation model SAM; through image preprocessing and enhancement, global multi-level semantic feature extraction, local fine segmentation driven by an attention thermodynamic diagram, bidirectional cross-modal feature fusion and a collaborative diagnosis decision network, bud root length level accurate classification and synchronous diagnosis of various growth anomalies are realized. The method has excellent recognition performance on small sample categories, is high in automation degree, strong in robustness and good in interpretability, and provides an efficient and reliable intelligent tool for rice seed vigor evaluation and sowing quality detection.

The invention relates to a financial risk control dynamic decision-making method, system and device and a medium. The method comprises the following steps: acquiring multi-source real-time risk characteristics and business environment indexes during customer transaction, and constructing an initial node structure associated with the characteristics through a graph neural network; effective connection is screened based on connection strength, and a dynamic routing scheme is generated by combining a high-risk scene optimization path priority; and utilizing reinforcement learning to iterate abnormal data flow analysis, and updating decision network configuration. The method comprises the following steps: extracting service demand indexes from configuration, generating an initial priority sequence through a graph neural network, harmonizing into a multi-target sequence by combining historical balance data, propagating risk characteristics in sequence to generate decision vectors, and verifying a matching degree to form a dynamic decision path. By adopting the method, the problem of insufficient scene identification of a traditional static model can be solved, the decision real-time performance is improved, the risk interception and business efficiency are balanced, a full-process dynamic risk control system is constructed, and efficient and adaptive risk decision support is provided for financial business.

The invention relates to the technical field of financial data intelligence, and discloses a financial personalized management classification system based on digital economy. According to the system, a dynamic mapping network is formed through weaving by constructing a user composite intention profile. And injecting a probe into a network node, performing compliance deconstruction and verification on a response sequence, driving behavior partition structure recombination, and performing re-calibration on the network according to the behavior partition structure recombination. And capturing an intention drift trajectory through a real-time data stream, dynamically and elastically telescoping network link weight, and finally fusing a strategy to complete management classification. According to the system, online driving optimization of data partitions in a compliance state and real-time dynamic shaping of a decision network by intention drift are realized, a closed-loop adaptive mechanism from bottom data to upper rules is formed, and classification accuracy, timeliness and compliance adaptability of financial personalized management in a dynamic environment are improved.

The invention relates to the field of path planning, in particular to a vehicle path planning method based on road network division and graph reinforcement learning, comprising the following steps: forming a directed graph according to a traffic road network, nodes in the graph representing traffic intersections, and edges representing connections between the intersections; dividing the real-time traffic flow data into a plurality of control regions according to the directed graph, assigning adjacent nodes among the regions as control nodes, and endowing each edge with a weight according to the real-time traffic flow data; inputting the directed graph into the graph convolutional network, extracting a road condition flow characteristic matrix, and combining the flow characteristic matrix, the current position matrix and the destination position matrix to obtain the current state of the intelligent agent; and the intelligent agent selects the advancing direction according to the decision network in the current state, updates the state according to the selected action, and optimizes the driving route. According to the method, efficient path planning can be realized in a complex and dynamically changing traffic environment, and the method is suitable for optimizing a vehicle driving route in real time.

This application relates to the field of logistics robot technology, and in particular to a logistics robot control method and system based on online transfer learning. It includes: collecting environmental state data through the logistics robot's sensor array, generating a state vector based on the environmental state data; inputting the state vector in parallel to two processing channels of a preset control algorithm, and outputting motion control commands for the logistics robot; updating the state vector based on the execution results of the motion control commands, generating learning signal parameters, and setting an optimization learning strategy for the control algorithm based on the learning signal parameters; and achieving functional decoupling of high-level decision-making and rapid adaptation by constructing a parallel dual-pathway architecture of a brain-decision network and a cerebellum-adaptive network. The brain-decision network outputs basic motion commands based on cross-environment task planning and strategy transfer; the cerebellum-adaptive network focuses on handling dynamic disturbances in the logistics scenario, providing real-time fine-tuning.

The invention belongs to the technical field of intelligent scheduling, and particularly relates to a heterogeneous cloud container cluster scheduling model training method and scheduling method based on topology awareness, and the training method comprises the steps: obtaining a cluster topological graph; obtaining a state vector time sequence of the nodes based on the node operation data, wherein each state vector comprises a performance index representing bottom layer resource contention; obtaining a queue state matrix; processing the state vector time sequence by using a feature extractor to obtain a node initial embedding vector; processing the initial embedded vectors of all the nodes by using a graph attention network to obtain a cluster global topology vector; obtaining a joint observation state based on the cluster global topology vector and the queue state matrix; inputting the joint observation state to the Actor-Critic network to generate a scheduling strategy; the accumulated rewards are maximized through a near-end strategy optimization algorithm, and network parameters are updated; and connecting the feature extractor, the graph attention network and the decision network to form a scheduling model. And the topological relation between the nodes is globally sensed, the interference attribute of the micro-architecture is captured, and accurate scheduling is realized.

The application provides a kind of mechanical arm cooperation grabbing method based on deep reinforcement learning, it is related to robot application technical field, to solve the problem of network instability in the method of multi-object grabbing in complex scene, low sample efficiency, unreasonable behavior.The method comprises: constructing a cooperative autonomous grabbing decision network model, obtaining the state-action value distribution of pushing and grabbing action by pixel-by-pixel prediction;Introduce object mask function, filter invalid area, reduce the influence of negative samples on model convergence speed;Design different behavior constraint strategy to suppress unreasonable behavior in network prediction, optimize the action selection of mechanical arm;Design a multi-element reward function, dynamically adjust the reward mechanism, guide the model to learn more action strategy that meets the task goal.The application is verified through simulation and real experiment environment, which can improve the task success rate and completion efficiency of the mechanical arm in unstructured scene, and has good robustness and applicability.

The application discloses a foreign language teaching effect evaluation method and system based on artificial intelligence, and the content comprises data collection, speech feature extraction, multi-modal fusion evaluation modeling and teaching effect evaluation. The application relates to the technical field of intelligent evaluation of teaching effect, and particularly discloses a foreign language teaching effect evaluation method and system based on artificial intelligence. The scheme constructs a pronunciation articulation composite index through formant difference analysis, calculates acoustic semantic coupling degree, and evaluates phoneme transfer stability, so that the articulation, accuracy and coherence of speech can be accurately quantified in a multi-noise environment. The scheme introduces modal attention, realizes bidirectional interaction of multi-modal information by using full-interconnection cross-modal cross-attention, and innovatively introduces a complementary gating mechanism to adaptively adjust the modal fusion ratio. Local scores are generated by combining a lightweight network, meta-decision network dynamically allocates weights, and finally, end-to-end optimization is realized by using a differentiable weighting strategy, so that the teaching effect evaluation of fineness and accuracy is realized.

The invention discloses a real estate business financial business risk control management method based on a big data model, and relates to the technical field of financial science and technology and risk management, and the method comprises the steps: collecting multi-dimensional original data, carrying out the fusion cleaning, and generating a structured risk control basic data set; extracting three feature sequences of application subject static portraits, real estate dynamic values and associated enterprise risks from the application subject static portraits and the real estate dynamic values; inputting the multi-source risk control feature fusion model into a pre-trained multi-source risk control feature fusion model for interactive modeling, and generating a comprehensive risk feature map; constructing a hierarchical risk decision network by using the map, and outputting a multi-dimensional vector containing credit and collateral risks and associated party conduction risk scores; and based on the vector, generating an adaptive credit line, a guarantee rate floating and a post-loan monitoring scheme through a dynamic strategy matching engine, and converting the adaptive credit line, the guarantee rate floating and the post-loan monitoring scheme into an executable instruction set to be issued and executed. According to the invention, the deep correlation analysis of the multi-source risk information and the automatic and accurate generation of the risk management strategy are realized.

The application discloses a network security situation awareness method based on artificial intelligence, comprising the following steps: collecting multi-source heterogeneous data to generate a standardized data set; decoupling space-time characteristics to separate space-time dimension characteristics; fusing space-time cross attention to output fused space-time characteristic vectors; constructing a causal reasoning engine to output a dynamic causal diagram and a counterfactual reasoning result set; constructing a dynamic risk propagation model to output a full asset risk value matrix and a risk propagation path diagram; generating a situation quantization matrix, constructing an adversarial training decision network, and outputting a defense strategy set verified by adversarial training; automatically generating a strategy; and man-machine collaborative verification closed loop. Through space-time characteristic decoupling and the causal reasoning engine, the application realizes dynamic reconstruction of a threat propagation path, reduces risk positioning error; in combination with the adversarial training decision network, the application makes the defense strategy have a reduced misjudgment rate in the simulation APT attack test.

This application belongs to the field of artificial intelligence and relates to a service provider allocation method, including: determining whether a service allocation request has been received; if so, parsing target service scenario information from the service allocation request; obtaining a target decision network corresponding to the target service scenario information; receiving a policy expectation value, input parameter values ​​corresponding to condition factors, and service provider parameters corresponding to the service provider; solving the policy on the target decision network based on the input parameter values ​​to obtain the optimal solution of the target policy state node corresponding to the policy expectation value; determining the target service provider based on the optimal solution and the service provider parameters, and allocating the target service provider to the target service scenario corresponding to the target service scenario information. This application also provides a service provider allocation device, computer equipment, and storage medium. Furthermore, this application relates to blockchain technology, and the target decision network can be stored in the blockchain. This application improves the processing efficiency and accuracy of service provider allocation.

The application relates to the technical field of component testing, and discloses an intelligent fault diagnosis method and system in an intelligent manufacturing process of a mine equipment. The method comprises the following steps: collecting running state information of the mine equipment in real time through a multi-source sensor; performing space coordinate registration after aligning vibration signals and acoustic emission signals through a time stamp to obtain a three-dimensional working condition atlas; decoupling features based on an equipment dynamics model to obtain a fault sensitive feature set; inputting the fault sensitive feature set into a multi-layer decision network to output a fault diagnosis result and a confidence score; constructing a fault cause-effect diagram based on pre-acquired design parameters and manufacturing process parameters; designing a correction scheme for abnormal process parameters in the fault cause-effect diagram, wherein the traceability confidence score of the abnormal process parameters is lower than a set threshold; and fusing the correction scheme, the fault diagnosis result and the three-dimensional working condition atlas to construct a visual diagnosis report. The application can improve the effect of intelligent fault diagnosis.

The invention relates to the technical field of food nondestructive testing, and discloses a multi-index meat freshness nondestructive testing method and system. According to the method, spectrum, image and gas sensing data of meat are collected and are separated into independent signal flows through signal decoupling. Performing targeted treatment: performing noise stripping and waveform locking on the spectrum to generate a characteristic spectrum; performing hierarchical analysis and morphological quantization on the image to generate a feature set; performing time sequence decomposition and gradient modeling on the odor to generate a feature sequence; and inputting the three types of features into a freshness collaborative decision network, performing cyclic comparison and evidence chain fusion through an interactive verification channel in the network, and outputting a freshness state report of a multi-index consistency conclusion. According to the method, deep collaborative analysis of multi-source heterogeneous information is realized, and the accuracy and stability of a detection result are improved.

The invention discloses a sample weight real-time adaptive statistical adjustment method for a classification model, and the method comprises the following steps: obtaining the sample data of a current training batch, inputting the sample data into a main classification model, processing the sample data through the main classification model, and generating prediction result data; the real-time statistical tracking module receives the prediction result data and the sample data, and updates a sample level statistical magnitude and a batch level statistical magnitude based on the prediction result data; the weight decision network receives the sample level statistics and the batch level statistics, and calculates a real-time weight value of each sample; updating the parameters of the main classification model by using a weighted loss function, and updating the parameters of the main classification model through a back propagation algorithm; updating parameters of the weight decision network according to updating feedback of the main classification model; therefore, the feature learning of minority class samples and difficult samples can be continuously optimized, and the classification balance among the classes is remarkably improved while the overall high accuracy is kept.

The invention belongs to the crossing field of artificial intelligence and production manufacturing, and discloses a flexible job shop scheduling method based on a Mama model and a cross attention mechanism, which comprises the following steps: constructing a Markov decision process model; serializing scheduling state characteristics; extracting features of the Mama model and the cross attention backbone network; decision network operation-machine pair selection; and optimizing an arrangement transformation strategy. Aiming at the limitation of feature extraction in a traditional scheduling method based on a graph neural network, the invention provides a scheduling method taking a Mama encoder and a cross attention decoder as a trunk, efficient sequence feature modeling and feature interaction are realized, and then a high-quality solution of flexible job shop scheduling is obtained. According to the flexible job shop scheduling method, efficient sequence modeling and feature interaction can be realized.

This invention discloses a method and system for detecting foreign objects in food based on dynamic routing with information entropy and graph convolution, belonging to the field of food detection technology. This invention generates a corresponding two-dimensional information entropy feature map by calculating the discrete data of a multi-scale basic feature tensor. This map is then input into a routing gating network to generate a spatial routing mask. The spatial routing mask is used to partition the multi-scale basic feature tensor into high-entropy and low-entropy feature streams, which are then subjected to topological and convolutional processing to generate first and second discriminative features. These features are then fused with the spatial routing mask to generate a full-graph fusion feature tensor, which is then input into an evidence decision network. The network outputs category judgment confidence and cognitive uncertainty, which are then compared to determine whether to generate a physical interception command and output a foreign object rejection result. This reduces the false detection rate of foreign objects in food detection and solves the problem of increased false detection rates in existing technologies.

The application discloses an intelligent decision method and system for warehouse operation process optimization, and relates to the technical field of intelligent warehouse management, which comprises the following steps: collecting warehouse multi-source heterogeneous data, and constructing a full-factor feature dataset; constructing a warehouse global coupling graph model, and calculating a global coupling weight matrix through a double-layer coupling mechanism of direct coupling attention and indirect chain coupling attention; constructing a coupling perception decision network, generating initial decisions for each operation link, and simultaneously calculating dynamic coupling constraint thresholds; performing global collaborative iterative optimization according to the global coupling weight matrix and the coupling constraint thresholds; and real-time monitoring of warehouse abnormal events triggers local re-optimization, and adjusting the global collaborative decision to obtain the final decision. The application aims to solve the problem that in the existing warehouse decision, each operation link is independently modeled and optimized, and there is a lack of global collaborative consideration of the strong coupling correlation among multiple links, which is easy to fall into local optimization.

This invention relates to the field of energy management technology in chemical industrial parks, and discloses an energy scheduling method for chemical industrial parks based on heterogeneous graph reinforcement learning. The method includes: acquiring the operating status of physical equipment and external environmental data of the park system to generate the current state; constructing a node feature matrix and a heterogeneous adjacency matrix based on the fixed physical connections of the park system, and inputting them into a GNN encoder to generate a current state embedding vector, which is then input into an action decision network to generate a current composite action vector; applying the action vector to a physical simulation environment to acquire the interaction variables and carbon emissions of physical equipment within the park system, and calculating a comprehensive reward; inputting the state embedding vector and action vector into a value network, and combining the comprehensive reward, backpropagating to update the parameters of the GNN encoder, action policy network, and value network, and obtaining the optimal comprehensive reward through iterative optimization to achieve optimal energy scheduling in the park; this invention realizes intelligent scheduling and optimization of energy in chemical industrial parks.

The invention discloses a layered fine-grained image forgery detection method and a layered fine-grained image forgery detection system. And the authenticity of the input image is preliminarily judged through the authenticity judgment module. Constructing a multi-branch pixel-level counterfeit positioning network, and obtaining multi-scale feature representation from low resolution to high resolution through a feature extraction backbone network; a high-resolution forged mask is obtained through pixel-level positioning network prediction; and the decision network constructs a coarse-to-fine reasoning path according to the multi-level label system, sequentially completes forgery property judgment, forgery type judgment, method-level traceability and platform-level traceability, and outputs a conditional probability of a corresponding-level label as a traceability result. The method has a unified multi-level label system, the systematicness and the accuracy of detection are remarkably improved, the source and the position of counterfeiting are visually presented through mask positioning and hierarchical classification, and the interpretability is high.