330 results about "Mix network" patented technology

The invention discloses a rolling bearing fault diagnosis method based on an SE-ResNeXt and Informer hybrid network model, and relates to the field of industrial process control, spatial features of time series data are extracted through a plurality of convolution modules in the SE-ResNeXt, an SE module is added for adaptively adjusting the response intensity of each channel, and a fault diagnosis result is obtained. The design of a residual network is combined, so that the feature extraction process is more efficient, and fault diagnosis analysis is performed through an Informer model. In the training process of the model, cross entropy is adopted as a loss function, and optimization is carried out through an Adam optimizer, so that the prediction accuracy and stability are improved. Experimental results show that the method shows excellent fault diagnosis capability on a public data set CWRU, has good robustness and prediction precision, is significantly superior to a traditional deep learning model, and provides a new thought for state monitoring and fault prediction under complex working conditions.

The invention discloses an industrial internet of things-oriented equipment digital twin operation and maintenance management system, and relates to the technical field of equipment management. The system comprises a data acquisition and preprocessing module, a digital twin model construction module, a data transmission and storage module, a state monitoring and fault diagnosis module, an operation and maintenance decision and optimization module and a visual interaction module. The data acquisition module adaptively acquires data through a sensor and preprocesses the data; the model construction adopts multi-scale and multi-model fusion; a hybrid network architecture and an encryption technology are used for transmission and storage; performing feature fusion and transfer learning for monitoring diagnosis; reinforcement learning and multi-agent collaboration are used for decision optimization; visual interaction supports VR / AR fusion. According to the invention, multiple modules work cooperatively, accurate acquisition, efficient transmission and storage of data are guaranteed, and accurate fault diagnosis and scientific operation and maintenance decision are realized through an advanced algorithm; the operation experience is improved through visual interaction; the method also has energy consumption optimization and supply chain cooperation capabilities, and can improve the operation and maintenance efficiency of industrial equipment and enterprise benefits.

The invention provides an electric energy load real-time data acquisition and analysis method based on a low-cost scheme, and the method comprises the following steps: constructing a distributed data acquisition network, employing a combined hardware architecture of a current / voltage sensor and a low-power-consumption MCU, and achieving the data acquisition optimization through a dynamic sampling rate adjustment mechanism, when the load fluctuation exceeds a set threshold value, the sampling rate is automatically increased, when the load is stable, the sampling rate is reduced to a calibration sampling rate, and a Kalman filtering algorithm is adopted to carry out real-time data preprocessing; a three-level data transmission system is established, the three-level data transmission system comprises an edge acquisition node layer, a convergence gateway layer and a cloud processing layer, an optimized AODV routing protocol is adopted among nodes to construct a star-shaped and net-shaped hybrid network topology, and an optimal transmission path is dynamically selected according to the signal strength and the network congestion condition; a differentiated QoS transmission strategy is designed, and data is divided into three priorities of fault alarm data, real-time monitoring data and historical data.

The invention discloses an industrial equipment defect detection method based on multi-modal feature fusion and dynamic optimization, and belongs to the technical field of computer vision, and the method comprises the steps: 1, multi-modal industrial data collection: deploying multiple sensors in a production line, collecting data in multiple periods, constructing a defect-free and multi-type defect sample library, and carrying out multi-modal industrial data collection; a time-space aligned multi-modal label is marked; step 2, data enhancement and defect synthesis; step 3, multi-modal hybrid model training: constructing a hybrid network, and performing pre-training and fine tuning by using a dynamic loss function; step 4, edge end dynamic optimization and deployment: edge end reasoning is realized through dynamic knowledge distillation, and model fine tuning is automatically triggered when false detection and missing detection are found; and step 5, intelligent labeling and result visualization: a front-end interface displays a detection result in real time. The problem that a current target detection framework is not high in small target recognition accuracy and low in efficiency is solved, and the reliability of industrial equipment defect detection is improved.

The invention relates to the technical field of remote monitoring, and discloses an intelligent AI-based fault detection method and system for a high-low voltage power distribution cabinet, and the method comprises the steps: constructing an LSTM-Attention hybrid network of a multi-modal data flow, extracting the time sequence characteristics of the multi-modal data flow, extracting the non-stationary signal frequency domain characteristics of the multi-modal data flow, and carrying out the fault detection of the multi-modal data flow. The method comprises the following steps: constructing a fault knowledge graph of a high-low voltage power distribution cabinet, analyzing a topological correlation feature vector of the high-low voltage power distribution cabinet, training a pre-constructed power distribution cabinet fault global model to obtain a trained power distribution cabinet fault global model, and analyzing a fault probability matrix of the high-low voltage power distribution cabinet by using the trained power distribution cabinet fault global model. And establishing a digital twinborn body of the high-low voltage power distribution cabinet, analyzing a fault evolution path of the high-low voltage power distribution cabinet by using the digital twinborn body, and analyzing a fault detection report of the high-low voltage power distribution cabinet. The maintenance efficiency of the high-low voltage power distribution cabinet can be improved.

The invention discloses a task unloading and resource allocation collaborative optimization method for an industrial hybrid network, and belongs to the technical field of industrial communication networks. The method comprises the following steps of: firstly, acquiring multi-dimensional attributes of an industrial field terminal task and a network node state in all directions through intelligent super-sensitive sensing and quantum-level monitoring means; thirdly, constructing a dual-objective function joint optimization model dominated by task total execution time delay minimization and assisted by resource utilization rate maximization, integrating factors such as local queuing time delay and transmission and calculation time delay under the influence of a quantum state, and setting strict constraint conditions; then, an improved quantum heuristic distributed co-evolution algorithm is adopted for solving, and a fine operation and self-adaptive quantum dynamic parameter adjustment mechanism is designed to approach an optimal solution; and finally, allocating tasks and resources according to an optimization result, establishing a continuous monitoring system, and rapidly restarting an optimization process in case of an abnormality.

The invention provides a distributed large-scale anti-traceability elastic network intelligent routing method and system, belongs to the field of network communication and network security, and is suitable for intelligent routing decision optimization in a dynamic network environment. The method is based on a multi-agent reinforcement learning framework, network nodes are mapped into independent agents, and path traceability risks are blocked through local information constraints; dynamic feature aggregation of a neighbor link state is realized by adopting a lightweight graph attention network, and the local sensing efficiency of a large-scale network is improved; a QMIX algorithm is introduced, and network parameters are optimized by nonlinear fusion of a local Q value and global graph state representation through a hybrid network; and in combination with a self-adaptive exploration mechanism driven by action entropy, the sudden change scene strategy response capability is enhanced. According to the system, in military anonymous communication, dark network data transmission and cross-border sensitive services, the anti-traceability and transmission efficiency balance can be remarkably improved, the characteristics of high concealment, high elasticity and low resource consumption are achieved, and a systematic routing solution is provided for a dynamic network environment.

The invention discloses a building mixed purpose classification method and system fused with multi-modal data, and belongs to the technical field of remote sensing image data processing, and the method comprises the steps: obtaining the multi-modal data of a research region, and carrying out the preprocessing of the multi-modal data; constructing a hybrid network model based on the convolutional neural network, wherein the hybrid network model comprises a parallel branch network, a buffer area module and a predictor which are connected in sequence; and inputting the preprocessed multi-modal data into the hybrid network model, extracting a multi-scale feature map of the multi-modal data through the parallel branch network, and extracting semantic information of a main environment feature region around a building in the multi-scale feature map through the buffer region module, and outputting the probability that the building belongs to each category through the predictor. According to the invention, the method achieves the recognition of the mixed purpose of the building based on the multi-modal data, can make up for the defects existing in the classification research of urban buildings only depending on single-modal data, and improves the recognition accuracy.

The invention relates to the technical field of radio frequency power matching, and discloses a direct-current radio frequency hybrid high-power network matching optimization method. The method comprises the following steps: collecting current fluctuation data of a direct-current power supply module and power reflection data of a radio-frequency emission module, and generating a hybrid network impedance characteristic set; calling a power grid topology analysis engine to analyze the set to obtain an impedance track characteristic matrix; performing dynamic trajectory optimization on the matrix based on a preset frequency domain stability constraint rule to generate corrected impedance trajectory distribution; fusing the corrected track distribution and the radio frequency load dynamic response data, and constructing multi-band matching topological feature representation; performing parameter decoupling on the topological feature representation through a nonlinear matching mapping engine to generate an optimal matching parameter configuration table; and finally adjusting the voltage stabilization coefficient of the DC power supply module and the tuning capacitance value of the radio frequency emission module according to the configuration table. According to the method, efficient matching optimization of the hybrid network is realized, and the stability and adaptability of system operation are improved.

The invention discloses a stainless steel thin-wall pipe bending forming production line control method and system. The method comprises the steps that state feature vectors are generated through data collection; establishing a hybrid network model, and optimizing hyper-parameters of the hybrid network model by using an OOA eagle optimization algorithm to obtain an improved hybrid network model; inputting the state feature vector into an improved hybrid network model, and outputting predicted values of the springback value, the wrinkling probability and the ovality which are about to occur in the current bending section; inputting the predicted value into a multi-objective optimization algorithm, and performing back calculation in real time to obtain an optimal compensation parameter set by taking minimization of springback, wrinkling risk and ovality as optimization objectives; and the compensation parameter set is transmitted to a physical execution unit for production line control including bending control, core rod control, clamping control and feeding control. And the first-pass yield of bending forming and the production control efficiency are improved.

The invention discloses a vehicle-mounted video transmission hybrid topology network, and relates to the technical field of vehicle-mounted video communication. The vehicle-mounted video transmission hybrid topology network comprises a plurality of transmission nodes, each transmission node comprises at least one input port and at least one output port, and the plurality of transmission nodes form a hybrid network of a tree topology structure and a daisy chain topology structure; wherein the tree topology structure in the hybrid network comprises a root node and a plurality of subordinate sub-nodes, the daisy chain topology structure is a sub-branch of the tree topology structure, and the hybrid network comprises a main link and a standby link which are used for signal transmission. Through innovative designs such as hybrid topology, dynamic bandwidth allocation, data retransmission, redundant channels, routing address query and the like, high bandwidth, low delay and high reliability in the field of vehicle-mounted video transmission are realized.

The invention discloses a grinding wheel online state monitoring system and method based on multi-sensor fusion. The monitoring system is composed of an acquisition module, a data preprocessing module and a man-machine interaction module. Firstly, multi-source information collection is carried out through a sensor, then time-frequency domain analysis and wavelet transformation are carried out on the collected information in a targeted mode, and data feature values are extracted; according to the method, image characteristic values such as a Hilbert spectrum are obtained through Hilbert-Huang transform and fast Fourier transform, high-dimensional data are subjected to importance evaluation through a reelief algorithm to obtain multi-modal data with high correlation, a ResNet-DenseNet-LSTM hybrid network model is built, accurate monitoring of the online state of the grinding wheel is achieved through data driving, and the accuracy of the online state of the grinding wheel is improved. And finally, integrating a monitoring algorithm in an upper computer interface. According to the invention, sufficient monitoring precision and accuracy can be ensured, online monitoring is realized, and a complicated process of offline data processing is avoided.

The invention discloses a power load prediction method and system. The method comprises the following steps: acquiring historical power attribute data and constructing a time sequence; carrying out standardization processing on the sequence; performing multi-scale decomposition and reconstruction on the standardized sequence by using a wavelet transform convolution module, and extracting a reconstructed sequence fused with multi-scale features; and inputting the reconstructed sequence into an xLSTM-Informer hybrid network, capturing time sequence dependence characteristics through xLSTM, and outputting a load prediction result through introducing an Informer model of a probability sparse attention mechanism. According to the method, the problems of insufficient long sequence dependence capture, single multi-scale feature extraction and low calculation efficiency are effectively solved, and the precision and stability of long-time prediction are improved.

The invention relates to a lightweight low-delay continuous authentication method and system based on a linear attention hybrid network, and belongs to the technical field of information. The method specifically comprises the following steps: S1, a registration stage: collecting behavior data of a legal user and preprocessing the behavior data; training a lightweight hybrid feature extraction model, and establishing and storing a user behavior contour; s2, an authentication stage: collecting user behavior data in real time and preprocessing the user behavior data, extracting behavior characteristics and matching the behavior characteristics with stored behavior contours, and executing access control operation according to a matching result. According to the method, a global attention mechanism and a space-time attention mechanism are introduced, the long-term dependency relationship in a behavior biological feature sequence is fully captured, and the robustness of the model to noise and abnormal values is enhanced; through effective fusion of multi-sensor data and global features of different time points, high-precision legality verification of a user identity in a small authentication window is finally realized.

The invention discloses an intelligent pipe network leakage active prediction and early warning system based on multi-technology fusion, and the system comprises a multi-source heterogeneous data fusion collection module, a spatial-temporal feature depth extraction module, a degradation trend prediction and residual life evaluation module, and a multi-stage early warning and decision generation module. The multi-source heterogeneous data fusion acquisition module acquires and fuses ultrasonic guided wave signals, pressure flow time sequence data and environmental factor data; the spatial-temporal feature depth extraction module extracts spatial-temporal fusion features through continuous wavelet transform and a CNN-LSTM hybrid network; the degradation trend prediction and residual life evaluation module determines a degradation level, predicts residual life and quantifies a pipe explosion risk probability; the multi-stage early warning and decision generation module generates graded early warning signals and maintenance strategy suggestions, the technology crossing from post-event detection to pre-event prediction is realized, and the scientificity and refinement level of operation and maintenance management of a pipe network are effectively improved.

The invention discloses a port container automatic scheduling method based on multi-agent reinforcement learning, and the method comprises the steps: S1, building a corresponding relation between equipment and agents, and constructing a task set; s2, collecting operation state data, and constructing global and local state vectors; s3, generating a scheduling constraint vector, and cutting actions according to the resource, storage yard and path state to form a feasible action set; s4, on the basis of an improved QPLEX algorithm, constructing an individual value network containing a dump structure, and calculating an individual action value; s5, constructing a joint action value hybrid network, and mixing individual values according to the global state vector to form joint action values; s6, constructing a training sample, differentiating and aggregating instant and delayed return, and updating network parameters; and S7, during online scheduling, selecting an optimal action combination according to the combined action value, and generating and issuing a scheduling instruction. According to the invention, automatic collaborative scheduling of port container operation is realized.

The invention discloses a virtual power plant load regulation method and system based on a deep learning scheduling strategy, and relates to the technical field of power plant load regulation, and the method comprises the steps: obtaining multi-source heterogeneous data, and obtaining a feature tensor through preprocessing; the feature tensor is input into an LSTM-Transform hybrid network model, and a context code is output; inputting the context code into the time sequence convolutional network model, outputting an ultra-short-term prediction result, and further obtaining a short-term prediction result; constructing a multi-objective optimization model, and modeling constraint conditions; solving the multi-objective optimization model through a preset layered architecture to obtain an optimal scheduling strategy, and generating a scheduling instruction; and issuing the scheduling instruction to distributed resources in the virtual power plant, and executing and converting the scheduling instruction into an equipment action. The method solves the problems that in the prior art, the optimization target is single, the dynamic adaptive capacity is lacked, and the real-time fluctuation response speed of the power grid is limited to a certain extent.

The invention discloses a virtual-real fusion network communication performance real-time monitoring method, device and equipment and a medium, and belongs to the field of communication monitoring, and the method comprises the steps: monitoring a communication link performance test instruction issued by a user in real time; wherein the communication link comprises a real link, a virtual link and a virtual real link; when a first instruction of only performing performance test on one communication link is received, determining the type of the communication link according to the nodes at the two ends of the communication link, and selecting a corresponding test method to test the performance of the communication link according to the type of the communication link; and when a second instruction for testing the performance of all the communication links is received, testing the performance of all the real links and all the virtual links in parallel, and then testing the performance of all the virtual links in batches according to the path lengths of the virtual links. Therefore, by implementing the application, the problem of low communication performance monitoring efficiency under a complex virtual-real hybrid network structure can be solved.

The invention discloses a large-scale bulk cargo equipment motion control optimization method based on deep learning, and relates to the field of computer systems, deep learning, intelligent control and industrial automation based on specific calculation models. Aiming at the problems of incomplete environment perception, poor dynamic environment adaptability and the like in a bulk material equipment control system, the method comprises the following steps: firstly, acquiring data and preprocessing through a multi-source sensor; secondly, constructing a cross-modal attention fusion model to realize feature alignment; and finally, an LSTM-PPO hybrid network architecture is designed, an LSTM layer processes time sequence state characteristics, and a PPO algorithm realizes control strategy optimization. Compared with the prior art, the precision and controllability in motion control of traditional large bulk cargo equipment can be improved, the operation efficiency and robustness of the system can be improved more easily, and the method can be widely applied to the fields of logistics, bulk cargo loading and unloading, industrial intelligent manufacturing and production.

A secure container framework is disclosed for executing embedded AI micro-models in hardware-constrained or hybrid network environments. The system includes a secure execution container configured to manage AI micro-model lifecycle stages, enforce symbolic constraints, evaluate runtime telemetry, and optionally invoke fallback behaviors through alternate models or rule sequences. Each container includes cryptographically verifiable components such as policy maps, fallback subgraphs, and execution metadata. The invention supports mesh or non-mesh deployments, peer coordination, and operation on CPUs, GPUs, microcontrollers, or other equivalent or similar functionality hardware. The framework enables verifiable, autonomous, and policy-governed embedded AI operation.

The invention discloses a rotary machine fault diagnosis method based on PyramidNet and Transform, and relates to the field of fault diagnosis, and the method comprises the following steps: carrying out the normalization preprocessing and sliding window segmentation of an original vibration signal, and generating a time sequence data segment with a fixed length; carrying out discrete wavelet transform on the time sequence, carrying out multilayer decomposition by adopting a Daubechies 4 wavelet basis function, and converting a signal from a time domain to a frequency domain; a hybrid network model based on PyramidNet and Transform is constructed, local features are extracted by increasing the number of channels layer by layer, and channel attention and space attention optimization is carried out in combination with a CBAM module; performing time sequence modeling on the output features, and capturing a long-distance dependency relationship of a time sequence through a multi-head self-attention mechanism; according to the method, the extracted global features are classified, fault categories are output, feature extraction and fault classification are carried out by adopting an improved hybrid network architecture, the defects of a traditional method in feature extraction and global dependency modeling are overcome, and the detection precision and robustness of fault diagnosis are effectively improved.

The present invention proposes a method for predicting the operating status of an electric vehicle charging pile based on a hybrid network of CNN and LSTM, which solves the problems of low efficiency and low accuracy in fault detection of charging piles. First, by acquiring the relevant historical typical fault data, information parameters, operating environment and other operating status data of the charging pile, the influencing factors of the fault pile are analyzed and the key indicators of the operating status of the charging pile are determined. Then, the key indicators are quantitatively analyzed and processed, and the weights of the key indicators are determined using the improved hierarchical analysis method. A hybrid network architecture of CNN and LSTM is built, and the training samples are trained using the hybrid network of CNN and LSTM, and the relevant parameters are adjusted in combination with the test samples to improve the classification accuracy. The model of the present invention is highly efficient and practical.

The invention discloses a low-voltage ammeter fuzzy correction and detection method based on double-domain feature decoupling, and belongs to the crossing field of computer vision and power equipment intelligent detection. Aiming at the problems of dual-domain feature coupling interference, multi-task cooperative defects, calculation efficiency restriction and the like in the prior art, three aspects of innovation are provided: a dual-domain feature decoupling hybrid network (DDFDN) is constructed, local texture features are extracted through an asymmetric ConvNeXt encoder of a spatial domain branch, a frequency domain branch AMSA decoder is combined with frequency domain gating attention (FSAS) to maintain a global structure, and a multi-task cooperative algorithm is provided for solving the problems of dual-domain feature coupling interference, multi-task cooperative defects, calculation efficiency restriction and the like. Multi-scale feature fusion is realized by adopting a dynamic gating weight; designing a frequency domain constraint and semantic alignment mixed loss function, and performing joint optimization through multi-band L1 constraint and detection network feature similarity; a lightweight YOLO-Element detection head is developed, and a rotation sensitive convolution module and a frequency domain enhancement ROIAlign module are integrated. According to the method, the technical bottlenecks of a traditional method in the aspects of digital edge recovery, artifact suppression and multi-task cooperation are effectively solved, and a high-precision and low-delay solution is provided for intelligent inspection of the electric meter in a complex scene.

The invention discloses a multi-label website fingerprint identification method and system based on attribution analysis, and the method comprises the steps: firstly carrying out the model attribution analysis of a training sample, and constructing an average discrimination template of each website category; high contribution areas of all categories are extracted according to the average discrimination template, and a structured mask template used for guiding identification is formed; in the identification stage, sliding window matching is carried out on input hybrid network traffic, the position of a potential pseudo sub-stream is judged according to the similarity between a mask template and a window sub-sequence, and single-label identification is carried out on the extracted pseudo sub-stream, so that effective analysis of multi-label traffic is realized. The method has the advantages of being high in structure perception, high in template generalization and good in mixed flow adaptability, the website recognition capacity in a complex network environment can be remarkably improved, and the method is suitable for scenes such as encrypted communication monitoring and anonymous channel flow analysis.

The invention discloses an image reconstruction method and system based on a hybrid network framework, a terminal and a storage medium, and the method comprises the steps: carrying out the prediction of an original image, obtaining the prediction data, obtaining the quality loss and the physical loss of momentum conservation in a prediction process, comparing the prediction data with the original data, and obtaining a prediction result; obtaining a matching value between the two to construct a total loss function; and introducing an NTK matrix to draw up a proper learning rate, training a hybrid network framework by using the learning rate and a total loss function, and finally obtaining a DeepONet-NTK hybrid network. According to the method, DeepONet is used for learning a nonlinear mapping relation in a multi-dimensional function space, and the NTK theory is combined to optimize the training process and improve the convergence speed and stability, so that the signal-to-noise ratio of a reconstructed image is reduced, and high-dimensional semantic features are reserved.

The invention discloses an operation stage identification method, system and device based on a time-frequency hybrid network, the time-frequency hybrid network is provided, the time-frequency hybrid network comprises a prediction stage and a refining stage, the prediction stage and the refining stage each comprise a single-stage time sequence convolution module, a wavelet time sequence convolution module and a Fourier attention module, and the wavelet time sequence convolution module is used for carrying out time sequence convolution on the single-stage time sequence convolution module. According to the method, a large receptive field can be obtained at a small cost, the method has higher frequency domain feature extraction capability than a common single-stage time sequence convolution module, the single-stage time sequence convolution module and a wavelet time sequence convolution module are combined, and local feature details and filtering can be better obtained; the Fourier attention module is provided, the Fourier attention module is formed by connecting a Fourier filtering module and an attention module in series, the Fourier filtering module can filter out unimportant features, the subsequent attention module can better model the global range, and the recognition efficiency in the operation stage is effectively improved.

The invention discloses a multi-agent sliding mode safety control method based on dynamic event triggering under hybrid network attack, and the control method comprises the following steps: building a multi-agent system model comprising spoofing attack and DoS attack; in order to cope with hybrid network attacks, an integral sliding mode surface function is constructed, an online estimation method is provided, and an unknown network attack mode can be estimated in real time in the absence of prior information; a novel self-adaptive sliding mode controller is developed, and the accessibility of a preset sliding mode surface is ensured; by introducing an auxiliary variable, a dynamic and static event triggering mechanism in the framework is considered, so that the use of communication resources is minimized; a Lyapunov stability theory and a finite time method are utilized to analyze the convergence of the multi-agent system under the control strategy. According to the control method provided by the invention, the influence of spoofing attack and DoS attack on the system can be eliminated, so that the state of the multi-agent system is consistent within finite time.

The invention discloses a power load time sequence anomaly detection method and system based on a Mama and LSTM hybrid network, and belongs to the technical field of power system data analysis and artificial intelligence, and the method comprises the steps: inputting the preprocessed power load and related factor time sequence data into a Mama-LSTM hybrid encoder; performing time sequence data reconstruction and anomaly probability prediction in parallel by using depth features output by an encoder, and performing model training by jointly optimizing reconstruction error loss and anomaly detection loss; calculating a comprehensive abnormal score based on the trained model, and judging an abnormal point by adopting a dynamic threshold value; and outputting an anomaly detection result and providing an analysis report containing an unsupervised evaluation index and multi-dimensional visualization. Through deep series fusion of Mama and LSTM, long-term dependence and complex modes in a power load sequence are effectively captured, and the accuracy, robustness and interpretability of anomaly detection are significantly improved in combination with a joint training strategy and an unsupervised evaluation system of the system.

The invention belongs to the field of communication network fault detection, and particularly discloses an Ethernet physical layer line connection fault detection and positioning method and system, a storage medium and a product, and the method comprises the steps that a master station of the Ethernet sends Ethernet frames containing sequential addressing bits, each slave station processes the Ethernet frames in sequence according to a connection sequence, and the Ethernet frames are processed by the slave stations; meanwhile, sequentially addressing bits are progressively increased; each slave station detects a PHY chip clock frequency and a state register of the network port in real time; and when any slave station detects that the clock frequency of the PHY chip of the network port is reduced below a preset threshold value or the state register is link-down, judging that a disconnection fault occurs, returning the Ethernet frame to the master station from the network port of the corresponding slave station, and judging a disconnection position by the master station according to a sequential addressing bit of the received Ethernet frame. According to the method, the reliability and the response speed of disconnection detection are improved, disconnection detection can be triggered after the physical layer is disconnected under the condition of hybrid network topology communication, the disconnection position is positioned, and a network topology structure is reformed.

The invention provides an industrial multi-protocol heterogeneous real-time video collaborative analysis method. The method comprises the steps that a lightweight compiler is developed, an operator dependency graph analysis module is integrated, instruction-level parallelism of a neural network processing unit is predicted through a graph neural network, and optimization intermediate representation is dynamically generated; constructing a protocol classification graph neural network model, carrying out protocol type self-identification and streaming media metadata automatic extraction, and converting a heterogeneous video stream into a standardized frame sequence; designing a multi-agent reinforcement learning model, and generating an optimal scheduling strategy through offline simulation training by taking a neural network processing unit calculation unit utilization rate, a memory bandwidth occupancy rate and a task queue depth as state spaces; predicting a video key region by adopting a hybrid network, and performing resolution downsampling on a non-key region in combination with motion vector analysis; the hardware-level scheduling adopts time slice rotation preemptive scheduling, and time slices are dynamically distributed according to the priority of an algorithm.