553 results about "Network identification" patented technology

The invention provides a medical text big data intelligent labeling and knowledge graph construction method and system, and relates to the technical field of medical text processing, and the method comprises the steps: recognizing entities in a medical text through a pre-trained deep bidirectional coding model and a multi-level entity recognition network, and carrying out the normalization processing in combination with a medical knowledge base, according to the method, the semantic association degree between entities is calculated, a relation extraction result is optimized based on comparative learning, comprehensive relation representation is obtained through fusion of multi-granularity coding and a gating mechanism, the entity relation is determined in combination with an inference rule network, a medical knowledge graph is constructed based on an initial knowledge graph, and the accuracy of medical text entities and relation labeling can be effectively improved. And a more perfect medical knowledge graph is constructed, which is helpful for assisting medical diagnosis and decision making.

The invention discloses a scene interactive AI rehabilitation evaluation training and health monitoring system, and relates to the technical field of rehabilitation medical treatment and artificial intelligence, a semantic perception module is used for collecting and recognizing voice input, facial expressions, action tracks and eye movement paths of a user in a training process, and extracting context parameters; the knowledge-driven training generation module is used for calling a rehabilitation knowledge graph constructed by a graph neural network based on context parameters and individual training history, and generating a multi-path training scheme; training a feedback regulation engine, collecting posture offset, physiological stress and emotion feedback, and dynamically adjusting task difficulty, rhythm and prompt mode based on a dual-channel reinforcement learning model; the prediction module fuses training and monitoring data, and predicts a network identification function degradation risk through degradation driving; the cloud edge fusion platform is used for realizing task quick response and graph strategy iterative updating; according to the invention, the individuation, self-adaption and intelligent prediction capabilities of rehabilitation training are improved, and the rehabilitation effect and the system practicability are obviously optimized.

The invention provides a multi-target tracking detection method and system based on cascade matching and trajectory confirmation, and relates to the technical field of digital video monitoring and tracking, and the method comprises the steps: obtaining a personnel construction video, and decomposing a video frame into a plurality of images; identifying a target in the current frame of image based on an improved YOLOv5 network, obtaining a target detection frame, and obtaining a prediction trajectory frame of the current frame through a Kalman filtering algorithm based on the trajectory frame of the previous frame; performing multi-target tracking on the target detection frame and the prediction trajectory frame of the current frame by using a DeepSort algorithm, respectively obtaining motion features and appearance features of the target detection frame and the prediction trajectory frame, performing cascade matching on the motion features and the appearance features of the target detection frame and the prediction trajectory frame by using a Hungary algorithm, and performing matching optimization by using an IOU matching algorithm; and carrying out fusion updating on the successfully matched target detection frame and prediction trajectory frame by using a Kalman filtering algorithm, and replacing the original prediction trajectory frame to obtain a tracking result.

The invention relates to the field of low-altitude economy, in particular to a server communication optimization method and system based on low-altitude economy. The method comprises the following steps: monitoring an unmanned aerial vehicle communication relay node in a low-altitude field; carrying out node space position positioning calculation on the unmanned aerial vehicle communication relay nodes one by one, carrying out distributed relay topology fitting, and constructing a low-altitude relay communication network; collecting low-altitude real-time interaction information based on a low-altitude relay communication network; performing communication link bandwidth calculation on the low-altitude real-time interaction information, performing dynamic communication protocol modulation, and constructing a dynamic low-altitude adaptive communication protocol; identifying real-time communication state parameters of the relay nodes based on the low-altitude relay communication network; performing inter-node communication channel identification on the real-time communication state parameters of the relay nodes, performing optimal channel scheduling decision, and constructing an optimal channel scheduling strategy; current user communication data of the ground server is monitored. According to the invention, low-delay and high-efficiency server communication optimization is realized.

The invention discloses a dynamic interaction method based on cooperation of a multi-modal dynamic fusion large model and an intelligent agent. The method comprises the following steps: performing feature extraction on user voice information to obtain a voice coding vector, a text semantic vector and an emotion feature vector; performing dynamic weight feature fusion on the voice coding vector, the text semantic vector and the emotion feature vector through a multi-modal dynamic fusion large model to obtain a fusion feature vector; inputting the fusion feature vector into an intention-scene coupling network, and identifying to obtain a user intention label; and identifying according to the user behavior log to obtain a user portrait tag, inputting the user intention tag and the user portrait tag into an autonomous decision-making agent, generating a target decision-making action through a lightweight policy network, and then interacting with the user according to the target decision-making action. The intelligent interaction efficiency and accuracy of the customer service system are improved, the interaction experience of the user is also improved, and the method can be widely applied to the technical field of artificial intelligence.

The invention discloses a storage hard disk remote diagnosis system and method based on the Internet of Things, and relates to the technical field of health management of the Internet of Things and storage equipment. The system comprises a data fusion module, a causal analysis module, a risk analysis module and a map construction module. The data fusion module collects SMART parameters, IO operation time sequence data and environment data through Internet of Things equipment, dynamically distributes multi-source data weights by using an attention mechanism, and extracts hard disk health state features. And the causal analysis module is combined with the multi-scale time sequence convolutional network and the Bayesian causal network to identify periodic abnormal fluctuation and generate a fault root cause analysis report. The risk analysis module matches historical cases through federal incremental learning, calculates a hard disk fault risk index and generates an early warning signal. And the atlas construction module optimizes resource isolation, data migration and response paths according to the risk indexes, generates a visual operation and maintenance atlas, provides fault positioning, risk links and repair priorities, and improves the operation and maintenance management efficiency.

The invention provides an intelligent detection system for forging defects of a forge piece product, and relates to the technical field of industrial intelligent detection.The intelligent detection system comprises the steps that multi-angle images of a to-be-detected workpiece are collected and then spliced, and a complete surface expansion view of the product is generated; obtaining defect types of the defect candidate regions through region coordinates and region sizes of the defect candidates; obtaining a confidence value of a corresponding defect type judgment result; a lightweight deep network recognition model is called for secondary judgment, new sample data are generated in a manner of supporting manual annotation, and an equipment end is connected with a programmable controller for intelligent defect detection of the workpiece to be detected. According to the invention, the problems of incomplete defect coverage, failure to realize high-precision automatic identification of multiple types of defects and influence on the detection accuracy and the production efficiency caused by diversified and complex forging surface defects and limited image acquisition angles in the prior art can be solved, comprehensive acquisition and splicing of multi-angle images are realized, and the detection accuracy and the production efficiency are improved. The technical effect of improving the defect detection accuracy of the forge piece product is achieved.

The invention relates to the technical field of atmospheric pollutant detection, and discloses a method and a system for detecting excessive emission of atmospheric pollutants. The method comprises the following steps: acquiring pollutant concentration data of multiple monitoring points in a target area to form an original data set; abnormal value detection and correction are carried out on the data set, sensor fault outliers are eliminated, and a preprocessed data set is obtained; extracting concentration change trend characteristics in a preset time window of each monitoring point, and constructing a spatial-temporal characteristic matrix; inputting the matrix into a pollutant diffusion model, calculating a transmission path and strength between monitoring points, and generating a regional transmission network; identifying a potential source region of abnormal fluctuation of pollutant concentration based on a network, and marking the potential source region as a candidate region to be checked; performing multi-scale concentration gradient analysis on the candidate area, and determining a key monitoring area; arranging mobile equipment in the key monitoring area, and collecting high-precision component data; and comparing the data with a standard emission source feature library, matching emission source types of which the similarity exceeds a threshold value, judging whether the emission exceeds the standard or not, and generating a detection report.

The invention provides a thermal runaway prevention and heat dissipation optimization method and system based on a capacitor module, and relates to the technical field of capacitor module thermal management, and the method comprises the steps: extracting the multi-source operation data features of a single capacitor through an improved layered visual attention network, carrying out the data complementation through combining a bidirectional probability diffusion model, and obtaining a multi-source operation data feature of the single capacitor; and discovering a causal relationship between network identification data by using a causal structure, generating a performance analysis report, constructing a spatio-temporal dynamic graph neural network to predict temperature field distribution, performing fault diagnosis and uncertainty quantification in combination with a multi-task learning framework, further constructing a layered early warning mechanism, and performing early warning based on a layered reinforcement learning framework. And a dynamic prediction map and a performance analysis report are combined, a heat dissipation strategy is optimized, cooperative control of the heat dissipation units is realized, an optimal heat dissipation scheme is finally obtained, thermal runaway is effectively prevented, and the safety and reliability of the capacitor module are improved.

The invention relates to the technical field of e-commerce supervision and prediction, in particular to a cross-border e-commerce retail risk supervision method and system. Comprising the following steps: acquiring cross-border e-commerce multi-source e-commerce data in real time by using the Internet of Things and an API interface; identifying heterogeneous risk factors by using a deep belief migration network; constructing a risk evolution model and generating a dynamic graph based on the spatial state transition type Markov convolutional network; adopting an abnormal subspace clustering algorithm to carry out self-adaptive clustering and real-time grading early warning on the risk modes; and finally, combining a reinforcement learning algorithm, an intelligent decision and a dynamic optimization risk intervention strategy with resource allocation, and outputting a whole-process risk management and control scheme. The intelligent, automatic and precise levels of cross-border e-commerce retail risk monitoring, early warning and response are improved, risk disposal lag and resource waste are effectively reduced, and the industry supervision safety guarantee capability is enhanced.

The invention relates to the technical field of fracture network fracture water channel identification, and discloses an identification method for analyzing a dominant flow channel in a three-dimensional fracture network based on a topological network. According to the method, probability distribution parameters such as geometric occurrence, gap width and density of a rock mass multi-scale fracture system in a target area are obtained through field surveying and mapping and three-dimensional scanning, and a spatial topological structure of the three-dimensional fracture system is reconstructed by adopting Monte Carlo method simulation and discrete fracture network modeling technology; secondly, abstracting the fracture network into a three-dimensional topological graph model based on a graph theory principle, defining fracture center points and cross points as graph nodes, and converting fracture sections into weighted edges; the method breaks through the continuous medium hypothesis limitation of traditional seepage analysis, and has important engineering application value in the fields of deep geological energy storage reservoir seepage risk assessment, shale gas fracture network optimization design, rock slope stability analysis and the like; compared with a traditional numerical simulation method, the method has the technical advantages of being high in calculation efficiency, good in prediction precision, high in multi-scale applicability and the like.

The invention discloses a gas leakage detection method, a program product and equipment, and the method comprises the steps: obtaining a collected continuous frame infrared image, and obtaining a multi-frame visible light image corresponding to the continuous frame infrared image; forming infrared input data of a pre-trained gas leakage detection model based on the continuous frames of infrared images, and forming visible light input data of the pre-trained gas leakage detection model based on multiple frames of visible light images; based on the infrared input data and the visible light input data, a gas fusion feature map is obtained through a gas feature extraction network in the gas leakage detection model; and based on the gas fusion feature map, identifying gas leakage position data through a gas position identification network in the gas leakage detection model.

The invention provides an image-based construction site violation recording method and device. The method comprises the following steps: acquiring a real-time image of a construction area through a camera, and preprocessing, including denoising, contrast enhancement and size adjustment; using an improved YOLOv5 network to identify hand and head features, positioning a constructor, and extracting position information, state information and action information; based on the extracted information, judging whether the constructors have illegal behaviors or not; if yes, the system controls the camera to capture and store violation images. According to the invention, by optimizing the image acquisition and processing technology and combining with the improved YOLOv5 network, the efficiency and accuracy of violation monitoring on the construction site are significantly improved, and the problems of poor image quality, low recognition precision and low violation judgment accuracy in violation monitoring on the construction site in the prior art are solved.

The invention discloses a network security malicious traffic tracing method based on a generative adversarial network, which comprises the following steps: collecting network traffic data, selecting a target traffic sample, extracting a protocol behavior, a communication structure and a time evolution characteristic of the target traffic sample, and carrying out semantic coding; mapping the multiple types of features to a unified embedding space through a multi-view fusion mode; constructing a manifold set based on known malicious samples, and generating an intermediate state sample sequence; constructing an attack context and path disturbance information in combination with the target sample, and generating a guide vector; inputting the guide vector and the random variable into a generator, and outputting a potential malicious sample representation; identifying the authenticity of the sample and evaluating the path similarity by using a discrimination network; and finally, constructing an attack path graph according to the similarity and the spatial relationship, and calculating a shortest path sequence to a known malicious sample to realize traceability identification of potential malicious behaviors. According to the invention, tracking analysis of malicious traffic sources is realized by using the generative adversarial network.

The invention provides a control equipment networking communication method and device, equipment and a storage medium, and the method comprises the steps: carrying out the control function feature extraction of access equipment, obtaining a control type and a communication demand, and executing control cooperative perception processing to generate a control function descriptor; calculating a control criticality index based on the descriptor, executing control-oriented link establishment processing, and generating a control-sensitive routing table; classifying and marking the communication data to obtain service labels, executing control correlation route aggregation processing, and generating a dynamic route execution plan; time slot allocation of control period perception is carried out on network bandwidth resources, time slot division is established, data transmission time sequence control is executed, and networking data transmission between devices is achieved. According to the method, control function perception and control criticality calculation are introduced, so that the network identifies the importance degrees of different control data, and differential transmission of the control data is realized by controlling time slot distribution of correlation routing aggregation and control period synchronization.

The invention discloses a rice nitrogen response regulation and control network analysis and breeding target identification system and method based on multi-omics data. According to the system, organic combination of regulation and control network construction based on single or multiple varieties of materials, key transcription factor recognition and accurate positioning of regulation and control areas where transcription factors play roles is achieved through an expression-chromatin accessibility correlation research method, and cis-trans effect distinguishing of the regulation and control areas is achieved through a deep learning model. The method comprises the following steps: carrying out nitrogen starvation pretreatment on rice, then carrying out nitrogen resupply, collecting a root sample, and carrying out ATAC-seq and RNA-seq sequencing; an eCAAS method is adopted to construct a regulation and control network, and key transcription factors are identified and accurately positioned; the chromatin accessibility difference of different varieties is predicted through a deep learning model, the cis-action effect and the trans-action effect are distinguished, an upstream transcription factor target is provided for genes dominated by the trans-effect, and haplotype and editable regulatory region targets available for direct breeding are provided for genes dominated by the cis-effect.

The disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate. Embodiments disclose methods for managing a condition in a UE. The method includes initiating a timer for disaster wait range; in case that the UE is switched off while the timer is running, and that the UE is switched on after the UE is switched off, wherein a universal subscriber identity module (USIM) in the UE remains as same, and the UE selects a public land mobile network (PLMN) for a disaster roaming: identifying a first time remaining for the timer for timeout at switch off, identifying a second time related to a time between switching off and switching on, and determining whether to restart the timer based on the first time and the second time.

The invention discloses a bearing defect intelligent detection method and system based on deep learning, and relates to the field of bearing defect detection. Multi-modal data such as bearing vibration, acoustics, thermal imaging and the like are acquired by using various sensors, and a four-dimensional feature tensor is constructed through preprocessing such as noise reduction and feature extraction; features are fused through a reconfigurable multi-branch convolutional neural network, and defects are identified and a development trend is predicted in combination with a meta-learning twin network; a decision threshold is optimized by adopting a quantum heuristic algorithm, and multi-level early warning is realized; and continuous evolution of the model is completed through edge-cloud collaboration and federated learning, the functions of data calibration compensation, model dynamic optimization and the like are achieved, and efficient and accurate detection of bearing defects is achieved. The detection time is remarkably shortened, and the positioning precision is high; the novel defect response speed is high, and faults can be predicted in advance; system energy consumption is reduced, model updating is improved, stable operation of equipment is effectively guaranteed, and cost reduction and efficiency improvement of industrial intelligent operation and maintenance are facilitated.

The invention belongs to the crossing field of computer technology and business management, and discloses a purchase return risk early warning and management method and system based on multi-mode intelligent auditing, electronic equipment and a computer readable storage medium. The method comprises the following steps: segmenting a PDF invoice image through an improved Apache PDFBox analyzer; a CLIP model and an OCR technology are combined, a cross-modal attention layer is used for correcting an OCR text, a multi-modal consistency difference value is generated, and invoice authenticity is judged; when a false invoice is detected, generating a standardized risk event, and extracting supplier, purchaser and transaction edge data from the graph database to construct a time sequence sub-graph; and identifying an abnormal mode based on the graph neural network, verifying the risk by using a time sequence risk scoring model, and executing a treatment strategy. According to the method, through multi-modal fusion, dynamic graph analysis and time sequence modeling, the problems of low false invoice detection precision, long abnormal fund tracing time consumption, risk prediction lag and the like in traditional auditing are solved.

The invention provides a service session control method for a multi-dimensional identification system, which comprises the following steps of: dynamically generating unique multi-dimensional identification information including an equipment identification EID, a service identification SID and a network identification NID when a user terminal accesses for the first time or accesses a network again; encrypting the multi-dimensional identification information by adopting a Hash algorithm, performing digital signature on the encrypted multi-dimensional identification information, sending the encrypted multi-dimensional identification information to a multi-dimensional identification server to complete registration, and performing associative storage on the multi-dimensional identification information and the equipment capability information by the multi-dimensional identification server; establishing an end-to-end communication link; the dynamic management mechanism continuously maintains the session state, and tracks and updates the multi-dimensional identification information of the user terminal in real time; when the equipment signal intensity is lower than the safety threshold or the electric quantity is lower than the threshold, the user can complete equipment migration without perception; terminating the session and releasing resources; according to the method, the safety, the efficiency, the consistency, the flexibility and the expansibility are remarkably improved, and the complex and changeable communication requirements can be better met.

The invention discloses an organization internal communication efficiency improvement method based on social network analysis. The method comprises the following steps: S1, collecting internal communication data of an organization; s2, preprocessing and standardizing the collected data; s3, constructing a communication network graph, and identifying communication frequency and intensity; s4, identifying communication bottlenecks and key nodes in combination with a social physics model and a graph attention network; s5, introducing a network fluid dynamic model and a swarm intelligence method to optimize a communication strategy; s6, continuously optimizing the network structure based on an adaptive optimization method and a context awareness technology; and S7, generating personalized communication optimization suggestions, and dynamically displaying communication network changes. According to the invention, the communication efficiency and decision-making ability in the organization are effectively improved.

The invention discloses a GIS combined electric appliance operation on-line detection method, and relates to the technical field, and the method comprises the following steps: obtaining a partial discharge original time domain signal, carrying out the signal enhancement processing of the original time domain signal, extracting a weak discharge pulse signal, and generating an enhanced discharge pulse signal; constructing a holographic time-frequency matrix atlas according to the enhanced discharge pulse signals; the holographic time-frequency matrix atlas is used as an input item, and a weak discharge type is identified based on a weak discharge neural network identification model; based on the identification result and the sensor array data, acquiring space coordinates of a discharge source, and outputting a discharge type and a positioning result to the monitoring terminal; according to the method, the holographic time-frequency multi-scale analysis algorithm and the weak discharge neural network recognition mechanism are fused, so that the recognition capability, the anti-noise robustness and the positioning precision of the GIS combined electric appliance online detection system on weak discharge are remarkably improved.

Systems and methods for training an encoder-decoder model are disclosed. An example method includes receiving, over a communications network, a plurality of extraction model outputs from a corresponding plurality of extraction models, each extraction model output received from a corresponding extraction model and each extraction model output including a respective plurality of key-value pairs corresponding to extracted text from one or more training documents, receiving, over the communications network, character recognition data corresponding to the one or more training documents, receiving, over the communications network, ground truth key-value data corresponding to the one or more training documents, and training the encoder-decoder model based at least in part on the plurality of extraction model outputs, the character recognition data, and the ground truth key-value data, wherein the trained encoder-decoder model is configured to generate key-value pairs for subsequent outputs of the plurality of extraction models.

The invention relates to the field of data management, in particular to a block chain-based vehicle service data management method and system. The method comprises the following steps: acquiring full-cycle vehicle service behavior data and a vehicle service protocol log; performing vehicle digital identifier identification and one-by-one service behavior type analysis on the full-period vehicle service behavior data to obtain a service behavior module of each vehicle; service-driven intelligent contract editing is carried out on the vehicle service protocol log, distributed block chain storage processing is carried out on a service behavior module of each vehicle, and a vehicle service block chain network is constructed; identifying real-time vehicle service behavior data; and carrying out block chain module positioning on the vehicle service block chain network according to the real-time vehicle service behavior data, and carrying out instant automatic updating execution to obtain an instant updated block chain network. According to the invention, efficient and safe vehicle service data management is realized, and more personalized and intelligent vehicle services are provided for vehicle owners.

An unmanned aerial vehicle (UAV) identification method based on blind source separation and deep learning is disclosed, in the method, firstly, the one-dimensional radar cross section millimeter wave data set of the UAV is acquired, and the mixed signal is obtained by mixing, and the improved FastICA algorithm is used to separate it. Secondly, the separated signal is converted into a two-dimensional image by data transformation, and the two-dimensional image is augmented, the obtained data set is divided into training set, validation set and test set. Thirdly, establish a UAV classification model based on Improved ResNet18 and train this model on the training set to achieve UAV classification. In the present invention, the training time of the network is not greatly increased while the network identification accuracy is improved, so that it can well complete the UAV type identification, and the design is more reasonable and effective.

The invention discloses a commercial password application security assessment evidence identification method based on image identification, relates to the technical field of image identification, and aims to assess employee advanced engineer certificates, extract information from multi-modal documents by using a Qwen-VL visual language model, generate vectors, label the vectors, convert the vectors into structured evidence nodes and store the structured evidence nodes in a graph database. Establishing various edges according to different associations, and constructing a preliminary evidence network; identifying certificate nodes, and clustering to form a certificate entity group; a correlation score is calculated by searching the support evidence through forward traceability, a correlation score is calculated by searching the contradiction evidence through reverse traceability, the credibility of the certificate entity group is dynamically adjusted, and the influence is propagated according to the updated credibility; and screening abnormal certificates lower than a threshold value, analyzing causes and contradictory points, listing abnormal information, visually displaying key evidences, the contradictory points and associated paths, and generating text description.

The invention provides a low-illumination small target image recognition method based on deep learning, and belongs to the field of image recognition. The method comprises the following steps: establishing a low-illumination small target data set, training a multi-scale enhancement model and a small target semantic segmentation model, enhancing a low-illumination image, sending the enhanced image into the semantic segmentation model, realizing segmentation identification of a small target, and positioning a target track according to a segmentation image of the small target. According to the invention, for a low-illumination small target image, the network identification precision is improved, and the probability of false detection and missing detection is effectively reduced; while the brightness and the contrast ratio of the low-light image are improved, the amplification of noise and the loss of details are inhibited, and the detail information is reserved, so that the accurate and stable detection and identification of the enhanced image are realized.

The invention discloses an abnormal transaction network identification method and device, electronic equipment and a storage medium, and relates to the field of artificial intelligence or other related technical fields, and the method comprises the steps: obtaining all transaction data sets of a financial institution in a target time period; a transaction network graph is constructed based on the transaction data set, and the transaction network graph is a network graph constructed with transaction subjects as nodes and transaction relations between the transaction subjects as boundaries; inputting the transaction network graph into an abnormal transaction network identification model, and outputting an abnormal transaction network sub-graph; and identifying the abnormal transaction network based on the abnormal transaction network sub-graph. According to the method and the device, the technical problems that a supervised learning method based on expert rules mainly focuses on analyzing transaction information of a single customer or a direct transaction partner of the single customer, and the recognition accuracy of large-scale abnormal transactions is relatively low in related technologies are solved.

The invention discloses a sugar net recognition system based on a block chain and federal learning, and the system comprises a data collection module which is used for collecting fundus image data and related medical data; the federal learning module is connected with the data acquisition module, the federal learning module comprises a plurality of participant nodes, and each participant node performs model training based on local data; the block chain module is connected with the federated learning module and used for storing model parameters, training data abstracts and data access record key information in the federated learning process; the sugar network identification module is used for receiving the global sugar network identification model trained by the federal learning module; and the result output and feedback module is connected with the sugar net identification module. According to the method, the data collaborative training model is integrated through federal learning, so that the sugar net recognition accuracy is improved, misdiagnosis and missed diagnosis are reduced, the block chain ensures the data to be safe and credible, the innovative formula assists optimization, the system application is wide, clinical and remote medical treatment is facilitated, and the sugar net diagnosis service quality is improved.

The invention discloses a smelting process monitoring system of a medium-frequency electric furnace and a process guidance method of the smelting process monitoring system. The system comprises an edge acquisition terminal and a cloud platform, the edge acquisition terminal comprises an intermediate frequency voltage transmitter used for acquiring a voltage signal at the output side of the intermediate frequency electric control cabinet and converting the voltage signal into a first analog signal; the intermediate-frequency current transmitter is used for collecting a current signal at the output side of the intermediate-frequency electric control cabinet and converting the current signal into a second analog signal; the analog quantity conversion module is used for converting the first analog signal and the second analog signal into digital signals; the intelligent thermodetector is used for measuring the temperature of the molten metal in real time and sending temperature data; the cloud platform is used for receiving and processing the digital signal and the temperature data and storing high-frequency sampling data; identifying a smelting stage through an LSTM network; power control is optimized on the basis of a Transform model and a GAN model; and integrating an expert knowledge base and providing process guidance. According to the system, comprehensive digitization and real-time accurate sensing of the smelting process are achieved; and the timeliness and scientificity of decision making are improved.