4662 results about "Network structure" patented technology

The invention relates to the technical field of computer vision and multi-modal learning, in particular to a weak supervision target detection method guided by cross-modal pseudo labels. According to the method, a labeled source domain data set is constructed to train an image classification teacher model, and a teacher-student network structure is constructed; clustering the regional features of the target domain image, allocating pseudo tags to each cluster by optimizing the allocation cost between the source domain category and the target domain cluster, and constructing a pseudo tag pool; and training a student model on the pseudo label pool for region feature detection of the target domain image. According to the method, a cross-modal attention mechanism is introduced, so that more accurate semantic alignment between a source category label and a target domain feature is realized; the stability of label distribution is improved by a structure keeping regular term; the generalization ability of the model is further enhanced by multiple rounds of pseudo-label confidence learning. The method can be widely applied to tasks such as target detection, cross-domain transfer learning and open world recognition, and efficient and accurate weak supervision target detection is realized.

The invention relates to the technical field of exception prediction, in particular to a storage cabinet exception trend prediction system based on time series data analysis, which comprises a state monitoring module, an interval sensing module, a path reconstruction module, a symptom activation module and an evolution prediction module. According to the method, the state vectors including the temperature, the voltage, the current and the door lock state are constructed and combined with the timestamp information to form the time sequence data sequence, and the dynamic expression mode of state change is established; a jump characteristic is analyzed by using a ratio of a time interval to a state change amplitude, a short-time disturbance path and a trend evolution path are distinguished by combining a jump rate statistical index, and an evolution activation signal is identified based on trend maintenance and non-fallback characteristics. On the basis, a neural network structure with long-time dependent learning ability is introduced to capture an aperiodic thermal anomaly trend in a state sequence, and the accuracy and timeliness of anomaly recognition are improved through multi-dimensional parameter cooperative processing and path construction logic.

The invention relates to the technical field of industrial chain monitoring and treatment, in particular to a monitoring method and system for industrial chain breakpoint treatment. The method comprises the steps that production, logistics, finance, policy and environment dynamic information is acquired through multi-source data, industrial chain comprehensive characteristics are generated through standardization, fractal dimension embedding expression and cross-dimension fusion, and historical trend dependency is introduced to enhance prospective prediction; a dynamic coupling network is constructed, and risk propagation intensity between nodes is quantified by using a dynamic edge weight, so that cross-level breakpoint propagation analysis is realized; risk indexes are calculated by fusing node features and a network structure, breakpoint candidate nodes are screened in combination with an adaptive threshold value, and a multi-step evolution trend is predicted by adopting a nonlinear propagation function and mapped into a multi-level early warning level. And generating a governance strategy according to the risk level, and evaluating the effect in real time and dynamically adjusting parameters through a closed-loop optimization mechanism. According to the invention, closed-loop management of risk identification, prediction and adaptive treatment is realized.

The invention relates to the technical field of communication, in particular to an underground power distribution room communication system and method based on heterogeneous network and multi-modal fusion, and the system comprises a three-dimensional heterogeneous network cooperation module, a cross-modal feature fusion module, a dynamic routing decision engine module, a self-adaptive interference suppression system and a cross-layer cooperation optimization module. The three-dimensional heterogeneous network cooperation module comprises a 4G / 5G wireless communication sub-layer, a power line carrier communication sub-layer and an edge computing management sub-layer; the cross-modal feature fusion module comprises a multi-modal encoder group, a shared feature projection layer, a twin network structure and a cross-modal attention fusion module; and the dynamic routing decision engine module comprises a state space monitoring unit, a routing optimization unit and a millisecond-level path switching strategy. Through the arrangement, the reliability, the real-time performance and the energy efficiency of communication of the underground power distribution room are systematically improved, and a high-robustness communication infrastructure is provided for an intelligent power grid.

The invention belongs to the technical field of polymer preparation, and particularly relates to a curved-surface polyaniline modified 3D printing polyurethane material and application thereof. According to the preparation method, an ice crystal hard template method and a polyether glycol soft template method are combined, so that the microstructure of polyaniline is adjusted, and the curved polyaniline with regular morphology is prepared. The curved-surface arc-shaped structure has a large specific surface area, can form more contact sites with resin when being directly melt-blended with the resin, is beneficial to construction of a conductive path, and is not easy to agglomerate due to the curved-surface structure. The polyaniline with the arc-shaped structure is of a three-dimensional space structure and is directly blended with the resin to form a randomly arranged conductive network structure, so that the composite material is not easy to break when being impacted, and can be better embedded between the resin to form a stress transfer channel, and a complex three-dimensional network structure is constructed; the mechanical property and the antistatic capability of the product are improved.

The invention discloses an automatic river network grading and river topological relation identification method based on a graph theory, and relates to the technical field of hydrological geographic information. The method comprises the following steps: acquiring and cleaning a vector river network, a key point location and DEM data of a target drainage basin; constructing an initial river network graph model based on the line element connection relationship; integrating DEM topographic evidence and graph theory connection features, constructing and solving a global potential energy field equation containing topographic driving and boundary constraint, and calculating flow potential energy attributes of nodes of the whole network to determine a flow relationship; based on the flow direction relation, identifying topology abnormal structures such as strong connectivity components in the network, and performing ring breaking processing by using direction confidence to generate a ring-free directed network structure; and performing river grade division based on a topology transfer rule, and associating the key point location to a river network skeleton. According to the method, through global potential energy field solving and topological optimization, the problems that the flow direction of the plain micro-geomorphic area is difficult to recognize and complex loops cannot be graded are solved, and automatic construction of the river network topology is achieved.

The invention relates to a power distribution network distributed energy bearing capacity assessment method and device. The method comprises the following steps: carrying out topology analysis on a network structure of a power distribution network to obtain an initial network topology model; obtaining a node dynamic feature data set based on the initial network topology model and the distributed energy access point data of the power distribution network; wherein the node dynamic characteristic data set comprises operation parameters of each node of the power distribution network in different load scenes; generating a parameter incidence matrix according to the node dynamic characteristic data set, and obtaining a bearing capacity reference model of the power distribution network according to the parameter incidence matrix and real-time data of the power distribution network in an operation state; wherein the parameter incidence matrix is used for quantifying the coupling degree between the operation parameters; and obtaining a risk distribution mapping graph according to the bearing capacity reference model, and identifying a potential overload area of the power distribution network based on the risk distribution mapping graph. According to the invention, power distribution network operation risk assessment can be accurately realized.

The invention discloses a cluster computing power energy efficiency perception scheduling and green computing system, which relates to the technical field of computers and comprises a multi-source energy efficiency perception and data acquisition module used for acquiring power consumption, utilization rate, temperature, cooling state, PUE index and environmental data of cluster nodes. According to the invention, through the multi-modal energy efficiency fusion sensing network and the multi-scale convolution and time sequence attention fusion network, multi-source heterogeneous energy efficiency data such as current, voltage, temperature, airflow and the like of a node level can be collected and fused in real time and with high precision, noise is effectively removed, abnormity self-correction is realized, the defect of energy efficiency sensing granularity in the prior art is made up, and the energy efficiency sensing precision is improved. And reliable input is provided for subsequent scheduling decisions. A cross-scale dynamic twinborn collaborative modeling mechanism is adopted, a physical information neural network and a computational fluid mechanics model are coupled, optimization is carried out through a generative adversarial network structure, and accurate prediction of a complex energy consumption evolution curve and a cooling flow field is achieved.

The invention discloses a lithium battery residual life prediction method and system and terminal equipment, and relates to the technical field of lithium battery health management. The method comprises the following steps: receiving a battery capacity attenuation sequence as an original input sequence, detecting and filtering abnormal data by adopting a 3 sigma criterion, and carrying out noise suppression processing on the battery capacity attenuation sequence through a Dropout mask; and carrying out normalization processing on the preprocessed battery capacity attenuation sequence, dividing the battery capacity attenuation sequence into a training set, a verification set and a test set through a sliding window algorithm, and generating a time sequence characteristic matrix and a corresponding residual service life label. According to the method, a neural network structure fusing trend prior perception and dynamic attention regulation is constructed, a multi-scale capacity modeling strategy is introduced to separate a degradation trend, fluctuation disturbance and high-frequency noise, and compared with a traditional time sequence neural network or a single attention model, pseudo fluctuation characteristics caused by capacity regeneration can be more effectively recognized, and the method is more efficient and more reliable. And the judgment accuracy of the model in a complex degradation scene is improved.

The invention provides an automatic early warning method for sudden weather in a target area, which belongs to the technical field of weather early warning, and comprises the following steps of: establishing a primary dense matrix by adopting adaptive filtering processing and a frequency domain signal separation technology, and generating a secondary dense matrix by applying a marine meteorological recognition model of a spiral progressive network structure; a dynamic statistical equation is used to calculate the physical coupling relationship of each parameter to establish a multi-scale weather process balance matrix, a maximum flow and minimum cut algorithm is used to optimize a weather system coupling relationship network to calculate a coupling degree matrix, and a dynamic threshold adjustment mechanism is established according to coupling strength parameters to adjust the early warning detection frequency. And based on a comparison result of the coupling degree moment order maximum characteristic value and a preset risk threshold value, establishing a grading early warning system and outputting a corresponding early warning signal to control an offshore oil and gas platform emergency response system. The technical problem that a multi-time scale weather process coupling relationship cannot be effectively processed is solved.

The invention relates to the technical field of neural network model construction, in particular to a lightweight neural network model construction method, which comprises the following steps: based on a target task data set, in a lightweight basic operator library comprising a depth separable convolution, an inverted residual structure and an attention mechanism module, constructing a lightweight neural network model; and searching and jointly optimizing network structure parameters and weight parameters through the differentiable neural architecture to obtain an initial lightweight network model. And deploying the initial model in a target hardware simulation environment, and generating a Pareto optimal model cluster through structural re-parameterization and hardware-aware progressive channel pruning iterative optimization by taking model precision, reasoning delay and memory occupancy as collaborative optimization targets. And selecting a reference student model from the clusters according to deployment constraints, constructing a distillation framework taking the initial model as a teacher model, and performing fine adjustment by adopting a mixed strategy fusing multi-dimensional distillation loss to obtain a final model. The model gives consideration to precision and efficiency, and the detection efficiency and the quality control level are improved.

The invention discloses a lightweight unmanned aerial vehicle target detection method based on multi-branch dynamic adaptive convolution, unmanned aerial vehicle target detection is performed by using a trained target detection network model, and the target detection network model is designed based on YOLOv11 and comprises a backbone network, a neck network and a detection head. The backbone network is constructed based on MobileNetV4 and introduces a multi-branch dynamic adaptive convolution module to replace part of original convolution modules, the neck network adopts a bidirectional feature pyramid network structure, on the basis of standard P3, P4 and P5 detection layers, a P2 layer is newly added to serve as a special small target detection head, an original C3K2 module is replaced with a lighter C2f module, and the C2f module is used as a small target detection head. The number of output channels of all convolution layers is uniformly compressed to 128. According to the unmanned aerial vehicle target detection method, the feature extraction capability of the model on the multi-scale and small-size unmanned aerial vehicle target can be effectively enhanced, the calculation overhead is low, and the detection precision and robustness are high.

The invention relates to the technical field of industrial robot control, and discloses an industrial robot trajectory optimization control method based on an intelligent algorithm. The method comprises the steps that joint position information, tool center point coordinates and a motion time sequence when a robot executes multiple tasks are collected and stored in a track database; after cleaning and screening data, extracting a feature set containing a path point sequence, speed distribution and an acceleration contour; constructing an intelligent optimization algorithm model of a neural network structure, and training by using the feature set to learn a trajectory optimization strategy; analyzing a target position coordinate and a motion constraint condition of the current task to obtain an initial track parameter; and inputting the initial parameters into the trained model, outputting an optimized track sequence containing a path point list and a speed curve, and generating a control instruction to drive the robot to move. The method can adapt to different tasks, improves track rationality and motion stability, and fits industrial production practice.

The invention belongs to the technical field of intelligent manufacturing, and particularly relates to a manufacturing quality prediction method and system based on a multi-mode sequential network and application, and the method comprises the steps: carrying out the preprocessing of the sequential data of a process manufacturing production line, obtaining a sample set, and carrying out the sequential division into a training set, a verification set and a test set; on the basis of the sample set, key features are enhanced through a frequency domain enhanced channel attention mechanism, a multi-period mode of a time sequence dependence and period sensing module is captured in combination with a multi-layer expansion convolutional network structure, and a multi-mode time sequence network model is constructed; and sequentially carrying out training set training, verification set parameter adjustment optimization and test set performance verification on the multi-modal sequential network model, and outputting a prediction result. According to the method, the deep dynamic association among the multivariable time series data can be mined, the accuracy and robustness of manufacturing quality prediction are improved, and an efficient and reliable technical scheme and an implementation path are provided for process industry quality control and intelligent optimization.

The invention discloses an aquaculture comprehensive guarantee method and system based on multi-modal data acquisition, and relates to the technical field of intelligent aquaculture, and the method comprises the steps: collecting aquaculture multi-modal data and an aquaculture image of an aquaculture region, carrying out the aquaculture feature capture of the aquaculture image, and outputting a visual feature group; inputting the visual feature group into a MobileNet lightweight model, and outputting a health state label and a confidence score; based on the health state label and the confidence score, outputting disease type data through a residual network structure optimized by transfer learning; constructing a graph neural network according to the disease type data, and generating a dynamic regulation and control scheme; and the control center executes the dynamic regulation and control scheme, carries out regulation and control effect detection and target comparison on the aquaculture area of the dynamic regulation and control scheme, judges whether the regulation and control effect is effective or not, and generates a feedback regulation instruction. According to the invention, through multi-modal data fusion and intelligent decision closed loop, dynamic and accurate guarantee of aquatic product health management is realized.

This application relates to the field of artificial intelligence technologies, and discloses a machine learning model search method, a related apparatus, and a device. In the method, before model search and quantization, a plurality of single bit models are generated based on a to-be-quantized model, and evaluation parameters of layer structures in the plurality of single bit models are obtained. Further, after a candidate model selected from a candidate set is trained and tested, to obtain a target model, a quantization weight of each layer structure in the target model may be determined based on a network structure of the target model and evaluation parameters of all layer structures in the target model, a layer structure with a maximum quantization weight in the target model is quantized, and a model obtained through quantization is added to the candidate set.

The invention discloses a diamond cutter path error compensation method based on cutter wear evolution, which comprises the following steps: S1, multi-modal sensor data acquisition and processing: in the machining process, key state information between a diamond cutter and a workpiece is acquired in real time through a multi-modal sensor; s2, tool wear modeling: key geometric parameters and evolution trend of tool wear are extracted from the collected and processed data, and tool wear modeling is carried out. And S3, artificial intelligence prediction of the abrasion loss: training a tool abrasion prediction model based on a deep learning network structure. And S4, tool path compensation is conducted, specifically, the predicted abrasion loss and the actual free-form surface model are combined, the tool path compensation amount and the compensated control point coordinates are calculated, a new machining path is generated, and dynamic compensation of tool cutting edge abrasion is achieved. And S5, system integration and closed-loop control: integrating the steps on processing equipment, and realizing real-time data interaction and synchronization through a unified data protocol.

The invention belongs to the technical field of artificial intelligence generation, and particularly relates to a Chinese painting line draft coloring method based on time sequence modeling and cross-modal fusion, which comprises the following steps of: acquiring a multi-modal Chinese painting data set, and taking a line draft image, text description and a reference image in the image data set as original input; after preprocessing, dividing into a training set and a test set; the line draft image, the text description and the reference image serve as input, a Chinese line draft coloring network model is trained, and a training model is obtained; the Chinese line drawing draft coloring network structure comprises a text time sequence modeling module and a multi-modal feature fusion module, the text time sequence modeling module extracts semantic features through a text encoder and introduces position coding and bidirectional LSTM to construct a time sequence relation between words, the multi-modal feature fusion module fuses text and image features, and the text and image features are integrated to form a multi-modal feature fusion model. Detail optimization and feature enhancement are carried out; through the multi-modal data set, time sequence modeling and cross-modal fusion, automatic high-quality coloring of the Chinese drawing draft is achieved.

The invention provides a landslide risk dynamic early warning method and system based on a Bayesian network, and belongs to the field of geological disaster intelligent prediction, and the method comprises the steps: constructing a disaster-inducing factor data set, and screening key factors based on a Pearson's correlation coefficient and an information gain method; an FP-Growth algorithm is further utilized to extract association rules among high-confidence factors, a Bayesian network structure is guided to be optimized, and a Bayesian network model with a causal relationship is constructed; the model supports an incremental learning mechanism based on a newly added landslide sample, can dynamically update a conditional probability table, and realizes landslide probability prediction and risk grade division in combination with a Bayesian forward reasoning result. The method has the advantages of being high in causal reasoning ability, high in model structure expression ability, excellent in prediction precision and capable of supporting real-time updating and risk partition, and is suitable for an intelligent risk assessment and early warning system for landslide disasters.

The invention discloses an oil reservoir production dynamic prediction method fusing discrete gradient information, and belongs to the technical field of oil reservoir development and artificial intelligence crossing, and the method comprises the steps: building a heterogeneous oil reservoir oil-water two-phase flow numerical simulation data set based on a numerical simulation method; designing a double-branch network structure and extracting spatial and physical characteristics of input field data in parallel, wherein the spatial and physical characteristics comprise a main characteristic coding branch and a differential operator branch; designing a backbone network to carry out deep nonlinear modeling; an efficient pressure and saturation field prediction neural network model is constructed based on a double-branch network structure and a backbone network, in a model training stage, spatial region observation points of part of time steps are used to participate in data item loss calculation, and meanwhile, physical control equation residuals are introduced into all time steps and a whole space to serve as physical loss items; a trained efficient pressure and saturation field prediction neural network model is obtained, and high-precision prediction of a full-time-sequence pressure field and a saturation field is achieved.

The invention discloses a water supply network global water quality prediction method based on a double flow-graph convolutional network, and belongs to the field of urban water supply. According to the method, a pipe network model of a directed graph is constructed, information flows in the upstream direction and the downstream direction are extracted respectively, feature learning is carried out through a parallel graph convolution module, a random mask mechanism is introduced during training to simulate sensor missing, and accurate prediction of the water quality concentration under the sparse monitoring condition is achieved. According to the double flow-graph convolution water quality prediction model provided by the invention, the practicability and coverage capability of the water quality prediction model under the condition of sparse monitoring data are remarkably improved; a double flow-graph convolution structure and a dynamic mask training mechanism are adopted, so that the robustness of the complexity of a pipe network is effectively enhanced; the constructed water quality prediction model has the characteristic of high response speed, can realize real-time prediction of water quality in combination with historical monitoring data and network structure information, is suitable for various scenes such as water quality monitoring, abnormal early warning and intelligent regulation and control, and is beneficial to improving the operation efficiency and management level of a water supply system.

The invention discloses a transformer fault diagnosis and root cause positioning method based on a space-time diagram neural network, and relates to the field of transformer fault diagnosis, and the method comprises the steps: S1, carrying out the preprocessing of the structure information and DGA time series data of a transformer, and obtaining a topological network structure diagram and a DGA time series; s2, obtaining a spatial vector based on a message passing mechanism of a graph convolutional network; s3, a time sequence vector is obtained in combination with a Transform encoder and multi-head self-attention; s4, obtaining space-time fusion features; s5, constructing a multi-task prediction head based on the space-time fusion feature, the fault type historical data and the fault root cause; and S6, carrying out fault detection and outputting a corresponding fault type and root cause positioning result. According to the application, the accuracy of fault type identification can be remarkably improved, and accurate positioning of the fault root cause is realized.

The invention discloses a watershed water resource supply and demand risk prediction and evaluation method based on a multilevel Bayesian causal network, and relates to the technical field of water resource supply and demand risk management.The watershed water resource supply and demand risk prediction and evaluation method comprises the steps that a water resource supply and demand risk diagnosis knowledge graph is constructed according to key variables and interrelations input by a user; constructing a multi-level Bayesian causal network structure; estimating conditional probability distribution among the nodes, and performing parameter learning and structure training on the Bayesian causal network; carrying out risk path identification through a reverse Bayesian reasoning method; outputting a posterior probability of water resource supply and demand risk prediction; based on a preset fuzzy character string matching algorithm, typical risk events and risk features are extracted; and according to the posterior probability and the risk characteristics, comprehensively evaluating the water resource supply and demand risk level. The method can improve the systematicness and scientificity of risk identification, is suitable for multi-link and multi-scale risk assessment and scheme comparison and selection in a complex drainage basin, and has high practical value and popularization prospect.

The invention relates to the technical field of artificial intelligence, in particular to a deep learning and reinforcement learning system and method for personalized recommendation, and the system comprises a user operation data storage module which stores historical operation data, a preprocessing module which processes the historical operation data, an embedding module which converts the processed data into low-dimensional vectors of a user and an article, and a processing module which processes the low-dimensional vectors of the user and the article. The reinforcement learning decision module is a core, a dual-network structure unit constructs a target and action network, a state expression unit constructs a hierarchical state, a reward calculation unit calculates a comprehensive reward, an experience playback unit stores and samples experience, a strategy selection unit adjusts exploration and utilization balance, and a data input module extracts real-time behavior characteristics and inputs the real-time behavior characteristics into a model. And the score feedback module collects user feedback, constructs a reward function and is used for model updating, so that accurate recommendation is realized, and the training convergence speed and the model performance are improved through a dual-network asynchronous updating architecture and a priority experience playback mechanism.

The invention discloses a voice decoding method, system and device based on electroencephalogram signals and a medium, and relates to the technical field of electroencephalogram signal processing.The method comprises the steps that reading electroencephalogram signals and voice signals in the reading process of a to-be-tested person and imagination electroencephalogram signals in the imagination reading process of the to-be-tested person are collected; inputting the reading electroencephalogram signals and the voice signals into the DRCL to generate electroencephalogram characteristics containing voice information; training the DBM by using the electroencephalogram characteristics as input and using the voice signals as output, and adjusting the DBM by using imaginary electroencephalogram signals to construct a voice synthesizer; a mapping relation between the imaginary electroencephalogram signals and the voice signals is generated through the voice synthesizer, and decoding from the electroencephalogram signals to the voice signals is completed; according to the method, a deep representation correlation learning method is provided, potential correlation between electroencephalogram and voice signals can be deeply mined through a multi-layer network structure, a complex mode which is difficult to recognize by a traditional model is captured, and the voice decoding process is more accurate.

The invention discloses an unmanned aerial vehicle communication data feature extraction method and an unknown type intrusion detection method. Physical layer information and network layer information of an unmanned aerial vehicle are acquired; wherein the physical layer information comprises the position, attitude, height and command type of the unmanned aerial vehicle, and the network layer information comprises an IP address, a port number, a protocol type and packet metadata; constructing an unmanned aerial vehicle state feature vector according to the physical layer information and the network layer information of the unmanned aerial vehicle; based on the trained CNN branch network, extracting communication data features used for intrusion detection in the unmanned aerial vehicle state feature vectors; a double-branch network structure of the convolutional neural network and the time sequence convolutional network is provided, a loss function is improved, an information entropy regularization item is introduced, efficient detection and distinguishing of unknown attack samples are achieved, the network structure has the advantages of being high in calculation efficiency and light in weight, and the method is suitable for large-scale popularization and application. The method is suitable for being deployed in resource-limited edge computing scenes such as unmanned aerial vehicles.

The invention discloses a risk assessment method and system based on a Bayesian network and an evidence theory, and relates to the technical field of risk intelligent assessment, and the method comprises the steps: determining an assessment dimension and an assessment index of a high-investment low-income risk of an educational institution in MOOC learning according to a human-cargo-field model and an industry report; constructing a MOOC learning risk assessment index system according to the assessment dimensions and the assessment indexes; constructing a Bayesian network structure according to the MOOC learning risk assessment index system; obtaining questionnaire data and expert opinions according to the MOOC learning risk assessment index system, and determining Bayesian network parameters based on the questionnaire data and the expert opinions; and inputting the Bayesian network parameters into the Bayesian network structure to obtain an assessment result of the high-investment low-income risk, the assessment result including a risk prediction result, a key risk factor and a sensitivity analysis result. According to the method, the high-investment and low-income risk in MOOC learning can be accurately evaluated.

The invention discloses a sleep classification method and system based on time-frequency combination, and belongs to the technical field of biomedical signal processing and artificial intelligence. In order to solve the problems that an existing method is high in manual dependence, insufficient in time-frequency feature fusion and low in long-time-sequence modeling efficiency, sleep stage classification is carried out mainly through automatic time-frequency feature extraction, a dynamic attention mechanism with memory enhancement and a lightweight multi-branch neural network structure. According to the method, efficient modeling and accurate classification of the multi-scale electroencephalogram signals can be achieved, the deep sleep recognition capability and the overall classification accuracy are improved, meanwhile, the model parameter quantity and the reasoning delay are remarkably reduced, and good real-time performance and clinical applicability are achieved.

The invention discloses a continuous time dynamics prediction method, system, terminal and medium fusing a diffusion model and a graph frequent differential equation, and relates to the technical field of dynamics prediction.The method comprises the steps that a multi-node time sequence is obtained, network structure inference is conducted on the multi-node time sequence through the diffusion model, and a potential graph structure between nodes is obtained; carrying out continuous time dynamic modeling by adopting a Scheng ordinary differential equation, and predicting a node state at any time point; diffusion reconstruction loss, dynamic prediction errors and structure sparsity constraints are constructed, total loss is established, network structure inference based on a diffusion model and dynamic modeling based on a Shenzheng differential equation are coupled based on the total loss, and collaborative training optimization is achieved. According to the method, the potential graph structure of the system can be stably recovered in a complex noise environment, high-precision and continuous prediction can be carried out on dynamic evolution of the potential graph structure, and the limitation that structure inference and continuous time modeling cannot be considered in the prior art is overcome.

The invention provides an intelligent identification and alarm method for respiratory suppression events in an anesthesia revival period, which comprises the following steps of: continuously acquiring high-frequency physiological data such as respiration, blood oxygen and electrocardio of a patient through multi-channel equipment, and establishing a dynamic causal network model fusing medical priori knowledge and clinical guidelines after standardized processing and feature extraction; a Granger causal test and a dynamic time warping algorithm are combined, a significant causal relationship among key physiological parameters is dynamically identified, a causal network structure is updated in real time, a causal analysis result is further input into a time sequence Bayesian network, calculation of a respiratory suppression event occurrence probability and reasoning path tracing are realized, and the probability of occurrence of a respiratory suppression event is calculated. According to the method and the system, the probability score is calculated, an interpretable medical logic evidence chain and thermodynamic diagram visualization are automatically generated, and if the probability score exceeds the limit, multi-mode alarm and data locking are synchronously triggered, so that the timeliness, intelligence and interpretability of respiratory suppression detection are improved, and clinical precise intervention is facilitated.