15783 results about "Data pre-processing" patented technology

The invention provides a building health monitoring and evaluation method and system based on a physical neural network, and the method comprises the steps: collecting the monitoring data of vibration, strain, displacement and the like of a building through a sensor network, and carrying out the data preprocessing based on wavelet transformation and a neural network; inputting the processed data into a physical constraint encoder, mapping the processed data to a physical space by adopting a Hamilton quantity encoding method, and constructing constraint conditions in combination with physical model parameters; performing feature matching by using a physical constraint neural network, and extracting key features; adopting a principal component analysis method to decompose early warning indexes, and constructing parameter mapping through a deep neural network; generating a hierarchical decision rule based on a decision tree algorithm; and carrying out constraint check and cost calculation on the candidate schemes, and finishing scheme sorting through an analytic hierarchy process. Physical knowledge is embedded into a deep learning framework, and the accuracy and reliability of monitoring evaluation are improved.

A modeling method for three-dimensional real-time measurement reconstruction, comprising: on the basis of a three-dimensional camera, a laser radar, and an infrared sensor device, generating a multi-source original data set; performing data preprocessing; performing feature extraction and data fusion; performing three-dimensional space construction; by means of a deep machine learning algorithm, performing real-time dynamic adjustment and optimization on a model; and, by adopting an advanced image processing algorithm, generating a high-precision three-dimensional model.

The present invention relates to the technical field of big data processing, and discloses a multi-source heterogeneous data fusion and processing method based on big data, including the following steps: data preprocessing, cleaning, complementing and feature extraction of multi-source heterogeneous data, providing standardized and high-quality input for subsequent processing; dynamic knowledge graph modeling, constructing a dynamic knowledge graph based on the semantic relationship of multi-source data, and generating a unified semantic representation; cross-modal alignment and deep fusion, semantically aligning multi-modal data, and generating a unified comprehensive representation through deep fusion of features; real-time stream processing, dynamically updating and processing real-time updated stream data to achieve real-time fusion of data; adaptive optimization, dynamically adjusting fusion strategy parameters, and improving the adaptability and robustness of the fusion strategy. The present invention realizes efficient fusion, real-time processing and semantic consistency expression of multi-source heterogeneous data through dynamic knowledge graph modeling, cross-modal alignment, real-time stream processing and adaptive optimization.

The invention belongs to the technical field of coal mine disaster early warning, and discloses a coal mine fire and gas disaster risk prediction method and system, and the method comprises the steps: collecting multi-source heterogeneous data collected at a mine site, and carrying out the data preprocessing and conversion; for the preprocessed image and video data, a Video Swin Transform model is adopted to extract spatial and temporal features, and a cross-modal association fusion Transform model is adopted to realize multi-modal feature unified fusion; mapping the fused multi-modal feature representation to a disaster-causing factor index space, calculating a risk assessment index containing a secondary interaction item, and dividing risk levels; and generating a disaster early warning strategy according to risk level analysis, issuing an early warning notice, performing regular maintenance and updating, and outputting a decision report. According to the method, unified processing and fusion representation of multi-source heterogeneous data can be realized, complex association among disaster-causing factors is accurately captured, the risk index with physical significance and engineering reference value is calculated, and high-quality decision support is provided for prevention and control of coal mine fire and gas disasters.

The invention relates to the technical field of marine environment monitoring, in particular to a time-space diagram neural network anomaly detection method and system for marine ecology. The method comprises the following steps: acquiring ocean monitoring data; performing data preprocessing on the acquired marine monitoring data; constructing a space-time diagram structure based on the preprocessed marine monitoring data; performing spatio-temporal feature extraction on the constructed spatio-temporal diagram structure by using an anomaly detection model; carrying out feature fusion on the extracted spatial-temporal features; and carrying out local anomaly and global anomaly scoring on the marine ecosystem based on a multi-scale anomaly scoring mechanism. According to the method, a unified data processing framework is designed, multi-source heterogeneous data such as satellite remote sensing data, water quality monitoring data and biological survey data can be processed at the same time, information contained in various kinds of data is fully utilized, and the comprehensiveness and accuracy of anomaly detection are improved. Through the construction of a space-time diagram structure, the incidence relation between different data sources is effectively captured, and more comprehensive feature expression is provided for anomaly detection.

The invention discloses an intelligent visualization and text association method for a multi-modal knowledge graph, and relates to the technical field of information processing. According to the intelligent visualization and text association method for the multi-modal knowledge graph, multi-modal data in an internal database of an enterprise is obtained, and a multi-modal feature vector is generated by using a data preprocessing and feature extraction technology; the feature vectors are mapped to a unified vector space through a cross-modal embedding algorithm, semantic alignment, association and fusion among multi-modal data are realized, and a complete multi-modal knowledge graph is constructed. Through bidirectional association and multi-modal joint query, the information acquisition efficiency is improved, and visual display of the knowledge graph is realized by using the graph neural network. Meanwhile, the knowledge graph is dynamically updated through a graph reasoning algorithm, and the timeliness and accuracy of the knowledge graph are ensured. A more efficient solution is provided for an enterprise in the aspects of management and application of multi-modal data, and the intelligent process of knowledge management is promoted.

The invention relates to the technical field of autonomous robot decision-making systems, in particular to an autonomous robot decision-making system and method based on multi-modal sensing fusion, and the system comprises a multi-modal data collection module which is used for obtaining multi-modal sensing data of an environment; the data preprocessing module is connected with the multi-modal data acquisition module and is used for preprocessing the multi-modal sensing data; the feature extraction module is connected with the data preprocessing module and used for extracting features from the preprocessed multi-modal sensing data; the dynamic space-time diagram construction module is connected with the feature extraction module and is used for constructing a dynamic space-time diagram based on the features; the multi-modal data fusion module is connected with the dynamic space-time diagram construction module and is used for fusing the multi-modal data in the dynamic space-time diagram; through adaptive time alignment and a cross-modal space attention mechanism, the system can effectively process space-time inconsistency of data of different sensors, and centimeter-level positioning precision is realized.

The invention provides a marine ranch water quality parameter real-time correction and compensation method and system for a multi-source sensor, and relates to the technical field of multi-source sensors, and the method comprises the steps: constructing a double-layer edge computing network, and connecting a sensor through a micro-service architecture to collect water quality data. And carrying out data preprocessing in combination with wavelet transform. And establishing a sensor digital twinborn model, and calculating the real-time credibility. Establishing a multi-dimensional sensor association network, optimizing a weight coefficient by adopting federal learning, and establishing a self-evolution correction parameter matrix; and fusing the sensor data by using a multi-task deep learning model to generate an initial correction value. And calculating a theoretical reference value through a space-time sequence prediction model. A compensation coefficient is adaptively adjusted by adopting a fuzzy decision tree, hierarchical water quality parameter correction is realized, and a closed-loop self-optimization intelligent correction system is formed through verification of a digital twin model. The accuracy and reliability of marine ranch water quality monitoring data are effectively improved.

The invention discloses an intelligent driving multi-sensor fusion data processing system, which belongs to the technical field of fusion data processing and comprises a data preprocessing unit, a data fusion unit and a data post-processing unit. The data preprocessing unit normalizes the original data of the laser radar, the camera, the millimeter wave radar and the ultrasonic sensor through a standardization, cleaning, noise reduction and calibration synchronization module; the feature level fusion module deeply digs the characteristics of each sensor, and fuses geometric, visual, motion and close-range obstacle features by using a deep neural network; the decision-making level fusion module generates a driving instruction through weighted voting and fuzzy logic reasoning through classification and situation evaluation; and the data post-processing unit combines the vehicle and road condition optimization instruction, evaluates the risk, and stores the data feedback optimization system. The system can improve the sensing precision and decision reliability, enhances the expansion adaptability of the system, and provides guarantee for safe and efficient operation of intelligent driving.

The invention relates to the technical field of geological risk monitoring and early warning in tunnel construction, in particular to a karst tunnel risk intelligent early warning method, device and equipment based on deep learning, and the method comprises the steps: obtaining multi-source coupling data; performing high-dimensional fusion on the processed multi-source coupling data by using a generative adversarial network, and reconstructing a high-precision three-dimensional model of the detection area; performing geological risk grade classification on the detection area to generate a corresponding risk distribution map; generating a dynamic risk evaluation index, and triggering an early warning signal when the risk level exceeds a preset threshold value; the method comprises the steps of multi-source data acquisition, data preprocessing, high-precision three-dimensional model reconstruction, risk grade classification, dynamic risk monitoring and early warning and the like, and can solve the problems of single risk detection means, insufficient evaluation precision and poor real-time performance in existing karst tunnel construction.

The invention discloses a deep learning-based deep fusion network production line fault prediction method, which comprises the following steps of: firstly, acquiring SCADA (Supervisory Control And Data Acquisition) system or sensor data of an automatic production line, and carrying out data preprocessing, feature extraction and fault prediction label construction; then a deep fusion network 1D-CNN-LSTM-Attention-DCE model combining a one-dimensional convolutional network, a long short-term memory neural network, a fusion attention mechanism and a dynamic cross entropy loss function DCE is established, the preprocessed data is input into the constructed model for training, and finally the trained model is used for fault prediction to obtain a fault prediction result. According to the method, a 1D-CNN-LSTM-Attention-DCE fault prediction model is provided, local features are extracted from time sequence data by using the 1D-CNN, then a long-time dependency relationship is captured through LSTM, and finally weight distribution of key features is enhanced through an attention mechanism, so that the processing capability of the fault prediction model on complex time sequence data is comprehensively improved, and the fault prediction efficiency is improved. And the fault of the production line can be accurately predicted within a certain time length range.

The invention belongs to the technical field of data transmission, and particularly relates to an SD-WAN low-delay data transmission method and system based on edge computing, a plurality of edge computing nodes are deployed in an SD-WAN architecture, and the edge computing nodes are distributed on a network edge side, close to terminal equipment or a branch mechanism and have data preprocessing and local computing capabilities; monitoring network state parameters of each link in the SD-WAN in real time, wherein the network state parameters comprise delay, bandwidth, packet loss rate and computing resource utilization rate of edge nodes; routing to-be-transmitted data to the target edge node, and if the node has local processing capability, executing data preprocessing or caching operation; otherwise, forwarding the data to an adjacent edge node or a cloud data center; a distributed cache strategy is adopted, when network congestion is detected according to data access frequency and timeliness requirements, non-real-time data is temporarily stored in a local cache, and asynchronous transmission is executed after a link is recovered, so that the method has the effect of providing higher-quality and more reliable network service for a user.

The invention discloses a knowledge graph construction method based on a fine-tuning large language model, and the method comprises the following steps: S1), data preprocessing: S2), extracting knowledge points and relationships in a data source, and obtaining preprocessed knowledge points and relationships; s3) disambiguating knowledge points and relationships; and S4) constructing a knowledge graph, forming a dynamic extensible knowledge graph, regularly using the AI model to re-analyze the text data in the later use process, and updating the knowledge graph. According to the method, the knowledge points are automatically extracted by utilizing the knowledge point extraction fine tuning model to determine the entity and the relation edge, the knowledge points are further identified through multiple disambiguation means, and the complex relation between the knowledge points is constructed, so that the efficiency and the accuracy of data processing are effectively improved; and finally, storing the data subjected to ambiguity elimination processing into a graph database model to construct a knowledge graph, thereby realizing automatic construction and updating of the knowledge graph.

The invention provides an electric vehicle energy storage battery fault diagnosis and predictive maintenance control method and system, and relates to the technical field of fault diagnosis, and the method comprises the steps: collecting the real-time working condition data and environment parameters of an electric vehicle energy storage battery, and carrying out the data preprocessing through a cloud server; carrying out life prediction by utilizing a multi-task deep learning framework, and generating a comprehensive state evaluation report; constructing a multi-target maintenance decision optimization model based on the evaluation report, and predicting an optimal maintenance time window; historical maintenance data and fault diagnosis experience are fused, and a maintenance operation list is generated; and performing resource scheduling optimization through a mixed integer programming algorithm to form an optimal battery maintenance scheme.

The invention relates to the technical field of river pollution identification, in particular to a pollutant identification and early warning method and system for a river patrol pollution source, and the method comprises the steps: constructing a three-dimensional monitoring network system, so as to build a three-dimensional data collection architecture; establishing a space-time reference unified framework to realize synchronous time service of different monitoring nodes; deploying an edge computing node, and implementing data preprocessing at an equipment end of the monitoring node; constructing a pollutant feature library which comprises spectral features, water quality parameter correlation features and visual morphological features of river pollutants, and establishing a water quality parameter correlation model of organic pollutants and inorganic pollutants, which comprises a plurality of feature dimensions; and a self-adaptive threshold early warning mechanism is established, a space-time composite early warning model is constructed, and accurate positioning and hazard degree grading early warning of pollution events are realized. According to the invention, through the hierarchical fusion model of technology fusion, the precision and speed of identifying the river pollutants are improved.

The invention provides a power prediction precision improvement method and system for a wind and light station, and relates to the technical field of new energy power generation, and the method comprises the steps: obtaining the historical power, weather and equipment operation data of the wind and light station through a data preprocessing module, and carrying out the time sequence decomposition, spatial interpolation, abnormal value detection and correction, and obtaining the fusion feature data. And constructing a deep learning prediction model based on the fused feature data, extracting spatial features by using a convolutional neural network, processing the spatial features by using a long-short-term memory network to obtain an initial prediction result, introducing an attention mechanism to perform weight allocation to obtain weighted features, and combining to obtain power prediction data. And judging the weather type according to the real-time meteorological condition, and correcting by adopting the matched correction model when the extreme weather is identified. And finally, an error compensation model is constructed based on historical prediction errors for compensation, a final power prediction result is obtained, error correction is performed according to actual power data, and compensation model parameters are updated.

The invention discloses a remote sensing image semantic segmentation method based on Mama and Transform architecture fusion. The method comprises the following steps: S1, data preprocessing; s2, pre-training the model by using the MAE; s3, preliminary feature extraction: performing preliminary feature extraction on the image subjected to MAE processing through linear projection, a convolutional layer and a nonlinear activation function to obtain a low-dimensional feature map; s4, CFF feature fusion: fusing the low-dimensional feature map and the image after MAE processing by using a CFF module; s5, feature enhancement of a Mamb-T module: inputting the feature map of the fused image into the Mamb-T module for operation to obtain further feature enhancement of the image; s6, multi-scale feature extraction; s7, a segmentation module, wherein the feature map is input to a Mask2Former segmentation module for final image segmentation; and S8, outputting and post-processing. According to the method, local and global features in the image can be captured, and the segmentation problem under a small water body and a complex background is effectively solved through combination of a self-adaptive attention mechanism and convolution operation.

The invention discloses a precise numerical control machine tool operation state intelligent control method and system, and particularly relates to the technical field of numerical control machine tools. Multi-dimensional operation data, including working temperature, load, current, pressure, cutting force and tool wear data, of a machine tool are collected in real time, a Z-score algorithm is adopted for data preprocessing, abnormal values are extracted, a prediction model is constructed and trained based on a machine learning algorithm, abnormal data are analyzed in real time, and the abnormal values of the machine tool are analyzed in real time. And accurately identifying the operation state of the machine tool and evaluating the health condition. According to the health state evaluation result, an intelligent control strategy is automatically generated, and the feeding speed, the spindle rotating speed and the cutting depth of the machine tool are dynamically adjusted, so that the operation state of the machine tool is optimized, potential faults are found and dealt with in time, the machining quality is improved, the faults are reduced, the production efficiency and the equipment stability are improved, and the maintenance cost is reduced.

The invention relates to an intelligent industrial equipment state monitoring device and method, and the method comprises the steps: installing a vibration sensor, an acoustic sensor and an image sensor at key parts of industrial equipment, and comprehensively collecting the operation data of the equipment. In the data preprocessing stage, vibration and acoustic signals are filtered and denoised, and image data are denoised and enhanced. Extracting multi-dimensional characteristic parameters from three aspects of vibration, acoustics and images; a random forest algorithm is adopted for model training, and the method comprises the steps of data preparation, segmentation, decision tree construction, voting mechanism, majority voting, model verification and optimization, testing and the like. In the model identification stage, the real-time features are input into the trained model, and the equipment state is determined through decision tree prediction and voting summarization. And when an abnormal state is detected, setting a multi-level alarm response according to the probability, and generating a maintenance suggestion. According to the method, the whole-process intelligent monitoring of the industrial equipment is realized, and the fault identification accuracy and the maintenance efficiency are improved.

The invention provides a livestock breeding risk intelligent assessment method and system based on multi-source data fusion, and the method comprises the steps: collecting livestock individual vital sign data, breeding environment parameters, management behavior data and risk-related historical data through Internet of Things equipment, and carrying out the data preprocessing to form a standardized multi-source data set; extracting risk features of individual, group and environment levels based on the data set, and fusing the risk features to form a multi-dimensional risk feature library; utilizing machine learning to construct a differentiated risk assessment model; analyzing the incidence relation between the risk factor and the actual event through the Bayesian network to calibrate the model; realizing livestock risk grade dynamic division and early warning based on the calibrated risk scoring system; and finally, generating intervention suggestions for risk quantitative evaluation, risk prevention and control decision and loss evaluation. According to the method, accurate evaluation of livestock breeding risks is realized, decision support is provided for breeding safety management, and the method has relatively high application value.

The invention discloses a power distribution cabinet fault detection method and system, and relates to the technical field of power equipment state monitoring, and the method comprises the steps: laying a multi-modal sensor array, constructing a multi-dimensional data collection network, collecting various types of sensing data in real time, and carrying out the data preprocessing; establishing a multi-modal data fusion model by using a deep learning algorithm, and forming a fault feature mapping relation through feature extraction, data correlation analysis and cross validation to realize collaborative perception of a data level; and constructing a multi-dimensional fault diagnosis decision tree based on a collaborative perception result, accurately positioning a specific device position where a fault occurs in combination with weight distribution and confidence evaluation of data of each sensor, and giving out evaluation results of a fault type and severity. The system effectively reduces the equipment failure rate, prolongs the service life, reduces the maintenance cost, improves the power supply reliability, promotes the development of an electric power system towards the intelligent operation and maintenance direction, and is of great significance to the improvement of the operation reliability and maintenance efficiency of a power distribution system.

The invention relates to the technical field of red tide anomaly detection, in particular to an ocean red tide anomaly detection method and system fusing multi-source remote sensing and a graph neural network. The method comprises the following steps: acquiring remote sensing image data, unmanned aerial vehicle image data and monitoring data of a monitoring point; performing data preprocessing on the acquired remote sensing image data and unmanned aerial vehicle image data; performing feature extraction and feature fusion on the remote sensing image and the unmanned aerial vehicle image to obtain remote sensing feature data; constructing a space-time diagram structure based on the monitoring data of the monitoring points to obtain diagram structure data; based on a cross-modal comparison self-supervised learning mechanism, carrying out consistency representation learning on a remote sensing feature mode and a graph structure feature mode; by introducing multi-source heterogeneous data and fusing a graph neural network modeling means, the limitation of a single data driving method in the aspects of coarse red tide recognition granularity, low space-time precision and the like is effectively broken through, and the meticulous property and global perception ability of red tide feature modeling are remarkably improved.

The invention discloses a government affair data sharing system based on a data security law risk control mode, and relates to the technical field of data management. According to the method, full-process closed-loop management from data classification and grading, differential desensitization, law and regulation compliance review, access authority precise control to behavior trajectory real-time analysis and risk trend prediction is realized, and the risk trend prediction is realized by introducing key technologies such as a context sensitive desensitization algorithm, legal knowledge graph reasoning, behavior portraits, semantic trajectory similarity scoring and risk rise rate monitoring. The method effectively improves the recognition and response capability of the hidden data re-recognition risk, and achieves the safe, controllable, compliant and ordered intelligent risk control management of government affair data in the sharing process.

The invention provides a data transmission method and transmission device applied to a vehicle-road cloud collaborative road network system, and the method comprises the steps: collecting vehicle operation state data through a vehicle-mounted terminal, collecting real-time road network topology data through road side equipment, and transmitting the data to a cloud server for data preprocessing; performing feature extraction on the preprocessed data based on a preset time-space correlation model, generating dynamic road network feature data, calling a cooperative transmission model to perform transmission path optimization on the dynamic road network feature data, generating a path optimization parameter set, and performing dynamic adjustment on a target transmission path according to the path optimization parameter set. Generating a self-adaptive transmission path adaptive to the current road network state; and distributing the target vehicle data packet to the roadside node and the cloud node through the self-adaptive transmission path, and carrying out retransmission scheduling on the data packet which is not successfully transmitted based on a feedback confirmation mechanism. According to the invention, the communication efficiency and the system robustness in a vehicle-road cloud collaborative scene can be improved.

The invention relates to the technical field of bridge structure safety monitoring and maintenance, and discloses a road bridge real-time monitoring and maintenance system based on intelligent perception, which comprises a data acquisition module used for acquiring multi-dimensional data of a bridge operation state through a deployed multi-source sensor, the multi-source sensor comprises a vibration sensor, a strain sensor, a temperature and humidity sensor and image acquisition equipment; the data preprocessing module is used for carrying out cleaning, synchronization and feature extraction on the collected multi-dimensional data to obtain dynamic features, environment features and visual features; and a data fusion module. According to the method, the comprehensiveness and accuracy of bridge monitoring are improved through multi-source data fusion, the reliability of an evaluation result is ensured in combination with physical constraint optimization, health trend prediction and early warning functions are realized based on a time sequence model, scientific decision support is provided for bridge maintenance, and the monitoring and management efficiency is improved.

The invention relates to the technical field of data analysis, provides an urban meteorological disaster data identification method and system based on deep reinforcement learning, and is used for effectively improving the model performance so as to enhance the accuracy and real-time performance of meteorological disaster monitoring. The method comprises the steps of obtaining a meteorological monitoring data set of a target city area, executing meteorological data preprocessing operation on the meteorological monitoring data set to obtain a preprocessed meteorological spatial-temporal feature set, calling a trained deep reinforcement learning recognition model, and performing dynamic disaster mode matching processing on the meteorological spatial-temporal feature set to obtain a dynamic disaster mode recognition model. And generating a meteorological disaster recognition result set of the target city region, generating a disaster coping strategy set according to the meteorological disaster recognition result set, and performing dynamic strategy optimization processing on the deep reinforcement learning recognition model based on the disaster coping strategy set to obtain an optimized deep reinforcement learning recognition model. And deploying the optimized deep reinforcement learning recognition model to a meteorological disaster monitoring system.

The invention discloses a three-degree-of-freedom parallel robot trajectory tracking method and system, and relates to the technical field of industry-university-research, and the method comprises the steps: obtaining motion data related to the execution trajectory of a three-degree-of-freedom parallel robot, and carrying out the data preprocessing; based on historical operation data, an intelligent prediction model is constructed to predict trajectory errors and is used for optimizing feedforward compensation of trajectory tracking control parameters; acquiring actual execution track information of the robot based on a visual sensor, and performing error calculation on the actual execution track information and an expected track to perform real-time track correction; dynamically adjusting control parameters in combination with error prediction data and visual feedback data; in the operation process of the robot, the control strategy is continuously optimized, and the intelligent prediction model is updated by using real-time data. Through the intelligent prediction model, visual feedback and adaptive control optimization, high-precision trajectory tracking, rapid error correction, energy consumption reduction and environmental adaptability enhancement are realized.

The invention discloses a live-action three-dimensional model construction method, system and device based on a digital city, and the method comprises the steps: obtaining the multi-source sensing data of a city, and carrying out the data preprocessing, and the multi-source sensing data comprises city image data, point cloud data and GIS data; converting the preprocessed multi-source sensing data in different formats into a preset standard format, and performing registration and fusion of the multi-source sensing data to obtain fused multi-source data; performing data optimization on the fused multi-source data, and constructing a city-level three-dimensional real scene model according to the optimized fused multi-source data and a preset three-dimensional space analysis tool; the preset three-dimensional space analysis tool provides space analysis and three-dimensional space visualization functions; according to the invention, efficient and accurate real-scene three-dimensional data of a city is provided, and three-dimensional and visual space information is provided.

The invention discloses a large language model knowledge base question answering system based on a multi-path fusion recall retrieval algorithm, and relates to the technical field of artificial intelligence application, and the system comprises the steps of S1, data collection, S2, data preprocessing, S3, knowledge base construction, S4, analysis and clustering, S5, knowledge recall, S6, weight adjustment, and S7, weight fusion. According to the method, a dynamic weight distribution module is arranged, so that various paths of recall weights such as keyword matching, semantic similarity and a knowledge graph can be flexibly adjusted according to different semantic scenes such as a technical scene and a product query scene, a recall result can be effectively optimized, knowledge related to problems can be accurately screened, and the method is high in practicability and high in practicability. And irrelevant information interference is reduced, the answer quality and the system efficiency are improved, and the user satisfaction is improved.

The invention discloses a coal mine hole fracture seepage analysis and prediction system based on a deep learning model, and relates to the technical field of coal mine safety. Comprising a geological data acquisition and three-dimensional model construction module, a data preprocessing and cleaning module, a deep learning feature extraction and training module, a graph neural network optimization and topology modeling module and a seepage prediction and risk assessment module, geological exploration data and hydrogeological information of a coal mine area are obtained, and a three-dimensional geological model containing pores and fractures is constructed. Through the deep learning method combining the convolutional neural network and the graph neural network, the accuracy of coal mine fracture seepage analysis is remarkably improved. The CNN extracts local features, the GNN optimizes a global topological structure, and the seepage prediction precision is improved. The model not only can accurately predict a seepage path, but also can evaluate safety risks, help mine managers to identify hidden dangers in advance, reduce water damage and gas accumulation accidents, and optimize mine safety management.