3353results about How to "Lower latency" patented technology

The invention provides a digital power grid power cloud resource real-time hierarchical distribution method, which comprises the steps of dividing power cloud resources into a plurality of sub-resource pools according to a physical architecture and a topological structure of a digital power grid, setting a special resource pool for key equipment and sensitive loads of the power grid, and dynamically adjusting network topology and a routing strategy. A low-delay and high-reliability network transmission channel is provided through a priority queue and a bandwidth guarantee mechanism; for each sub-resource pool, extracting periodicity, tendency and randomness characteristics of a load by analyzing historical load data and a current system state of power grid equipment, and updating a load prediction result in real time; after incremental resources are migrated, a power grid control request is routed to a corresponding computing node by adopting a consistent Hash algorithm in a sub-resource pool, load balancing is realized by dynamically adjusting the number and distribution of virtual nodes of a Hash ring, and weights of the virtual nodes are dynamically adjusted according to load optimization indexes of the nodes so as to adapt to power grid load changes.

The invention relates to the technical field of electric power design, in particular to an intelligent dispatching system of a transformer area power supply grid. According to the method, a self-adaptive sampling strategy of real-time load fluctuation and distributed energy output characteristics is adopted; when the system detects the rapid change or abnormal fluctuation of the load, the data acquisition frequency is automatically improved, so that the capture of key information is ensured; when the load is stable, the sampling frequency is reduced, and the collection of redundant data is reduced; adaptive sampling can dynamically adjust the acquisition frequency, it is guaranteed that important data are not lost at the critical moment, and meanwhile the system burden is reduced; a hierarchical data processing architecture is adopted, data processing is divided into an edge layer and a cloud layer, the edge layer is responsible for conducting primary processing on data collected in a transformer area, including denoising, feature extraction and the like, and the data quality is guaranteed; only the key data subjected to primary processing is uploaded to a cloud end for data analysis and decision optimization; therefore, the burden of data transmission and processing is reduced, and the real-time response capability of the scheduling system is improved.

The invention provides a real-time streaming media transmission method and system based on a cloud game and a storage medium, and the method comprises the steps: monitoring network state parameters in real time, and generating network state data; dynamically adjusting the resolution, the frame rate and the bit rate of the video stream based on the network state; a redundant data packet is inserted, and the redundancy proportion is adjusted according to the packet loss rate and the jitter, so that the data reliability is enhanced; dynamically adjusting the transmission rate by adopting a congestion control algorithm; dynamically selecting an optimal server node according to a network condition and a server load; audio and video processing tasks are shared through edge computing nodes, and the load and transmission delay of a main server are reduced; setting a self-adaptive decoding buffer area, dynamically adjusting the size of the buffer area according to a network state, and reducing playing lagging; through real-time monitoring, dynamic adjustment and edge collaboration, smoothness is guaranteed, reliability is improved, delay is reduced, efficient load distribution and user experience optimization are achieved, and the method is suitable for real-time streaming media transmission in a cloud game scene.

The invention belongs to the technical field of image classification neural networks, and discloses a lightweight image classification neural network architecture system based on knowledge distillation. The invention provides a lightweight image classification neural network architecture system based on knowledge distillation. The lightweight image classification neural network architecture system comprises a data set preprocessing module, a teacher model preprocessing module, a subnet model construction module, a personalized training module, a real loss calculation module, a distillation loss calculation module, a total loss calculation module and a student model construction module. The invention provides a lightweight neural network architecture system based on knowledge distillation, and designs a brand new training mode by using the superiority of knowledge distillation for teacher model weight extraction: the knowledge distillation of teacher and student models is combined with a deep compression method; the system for constructing the lightweight network can be used for training various complex image classification neural networks so as to improve the network performance.

The invention relates to the technical field of environmental sound event detection, in particular to an environmental sound event detection method based on multi-modal data fusion, which utilizes a microphone array, a camera and a sensor to synchronously acquire audio, video and environmental data to form a multi-modal data source. And the collected multi-modal data is preprocessed, so that the data quality and the accuracy of subsequent analysis are improved. And key features are extracted from the processed data, and the features are fused to form multi-modal features. And a deep learning model is constructed, and multi-modal features are used for training, so that a model capable of identifying an environmental sound event can be obtained. And performing optimization such as pruning, quantification and knowledge distillation on the trained model, and deploying the model on edge equipment to realize real-time preliminary feature extraction and event detection. The problems that an environment sound event detection method in the prior art is poor in robustness, low in real-time performance and limited in generalization ability are solved.

The invention discloses a data real-time analysis and decision-making system based on an edge intelligent Internet of Things, belongs to the technical field of Internet of Things, and aims to solve the technical problems of insufficient real-time performance, high delay, large bandwidth occupation and high dependence on a central server in data processing of Internet of Things equipment. Comprising an edge device layer, an edge computing node layer, a cloud edge collaboration module and a central server layer, each edge computing node is used for taking data after local analysis processing as input, performing prediction analysis through a local model, making a decision through a decision generation mechanism based on an analysis result, and performing cloud edge collaboration. Generating decision optimization through an adaptive decision optimization algorithm, and performing data exchange with the central server layer through a cloud edge collaboration module; and the central server layer is used for carrying out model updating based on the real situation and the analysis result corresponding to the analyzed and processed data uploaded by each edge computing node, and returning the updated model parameters to the edge computing node layer.

The invention discloses an unmanned aerial vehicle path planning method and system based on digital twinning. The method comprises the following steps: constructing a three-dimensional environment model based on first target environment data in first target multi-source environment perception data; constructing a digital twin based on the geometric model, the physical model, the control model, the target physical data hybrid dynamic model and the three-dimensional environment model; inputting the second target multi-source environment sensing data into the digital twin, and dynamically simulating the flight behavior of the physical unmanned aerial vehicle based on the second target multi-source environment sensing data to obtain simulated flight environment state data; and based on the simulated flight environment state data, performing path planning on the physical unmanned aerial vehicle in combination with a prediction control algorithm and a path planning algorithm to obtain a target flight path, transmitting the target flight path to the physical unmanned aerial vehicle, and controlling the flight of the physical unmanned aerial vehicle. The method can dynamically simulate the real-time change of the physical unmanned aerial vehicle, plan the optimal flight path and effectively avoid obstacles, and can be widely applied to the technical field of unmanned aerial vehicle path control.

The invention discloses a cloud computing storage, calculation and training resource allocation method based on reinforcement learning and dynamic resource scheduling. The method comprises the following steps: S1, establishing a unified resource scheduling framework; s2, constructing a state space and an action space; s3, generating a state representation for predicting future resource demands; s4, initializing a reward function of key performance index weighted comprehensive evaluation; s5, generating an initial storage resource, computing resource and training resource allocation strategy; s6, continuously adjusting a reward function of the reinforcement learning model based on the self-supervised learning feedback signal; s7, dynamically adjusting a distribution scheme of the storage resources, the computing resources and the training resources through quick response of the resource scheduling framework; s8, optimizing resource allocation among a plurality of cloud computing nodes or platforms; s9, automatically identifying different stages of the current task through a self-supervised learning model; and S10, finally generating an optimal allocation scheme. According to the staged resource scheduling method, the condition of overload or insufficiency of resources is effectively avoided, and the task execution time is greatly shortened.

The invention provides a central air-conditioning system real-time fault prediction method and system based on digital twinning, and the method comprises the steps: deploying various types of sensors for multi-dimensional detection, and collecting the data of the sensors; edge computing nodes are deployed around the central air conditioning unit; a digital twinborn model based on the central air conditioning unit is constructed according to the preprocessed data, a dynamic view of key performance indexes is provided, and the state of the central air conditioning unit is mapped and simulated in real time; designing a fault prediction model, and predicting and identifying faults; deploying the fault prediction model at an edge computing node and operating the fault prediction model, and performing real-time fault monitoring and diagnosis; when a fault is found, the AI-drive maintains decision support, and automatically adjusts a fault prediction model parameter or arranges a maintenance strategy; and constructing a learning loop based on performance feedback, and optimizing a fault prediction model and a maintenance strategy. Quantum calculation, the digital twinning technology and edge calculation are fused, and the operation efficiency and reliability of the central air conditioning system are remarkably improved.

The invention relates to the technical field of information synchronization, in particular to a cross-platform electronic communication synchronization system, which comprises a network construction and path optimization module for collecting basic network parameters of communication equipment of a plurality of platform networks, obtaining cross-platform network data according to the basic network parameters, selecting an optimal transmission path according to the cross-platform network data, and transmitting the optimal transmission path to a server; and generating real-time optimized path information. In the invention, the basic network parameters of the communication equipment of the plurality of platform networks are collected, the cross-platform network data are acquired, the edge weight is dynamically updated according to the delay and the bandwidth, and the optimal transmission path is selected based on the network state, so that the effect of efficiently utilizing network resources is achieved, and the real-time performance and the accuracy of synchronous data transmission are remarkably improved. The to-be-synchronized data is classified and subjected to priority setting, and the data processing sequence is adjusted in combination with the real-time network data, so that accurate scheduling of the synchronization task is realized, the delay and conflict problems in data transmission are reduced, and the efficiency and reliability of data synchronization are improved.

The invention relates to the technical field of communication networks, and discloses a service channel time slot allocation method, which comprises the following steps of: 1, acquiring network data in real time from network equipment, the network data comprising bandwidth utilization rate, delay, packet loss rate and service flow type; 2, predicting future network flow fluctuation through a long-short-term memory network model based on historical flow data and real-time network data; and step 3, dynamically adjusting the allocation of time slot resources according to the network flow fluctuation prediction result, and preferentially allocating the time slot resources for high-priority services. According to the method, the dynamic adjustment of the time slot resources is realized by combining the flow prediction of the LSTM model and the Q-learning reinforcement learning algorithm, so that the system can flexibly allocate the resources according to the actual network load condition, the resource waste is reduced, the utilization efficiency of the bandwidth is improved, and the utilization rate of the overall network resources is remarkably improved especially under the condition of high network load.

The invention discloses an intelligent advertisement putting processing method and system based on user data, and relates to the technical field of data processing, and the method comprises the steps: putting an advertisement in an advertisement pool based on an edge device, collecting the video information during the advertisement playing period through the edge device, analyzing the stored video information, extracting the user feature information, and storing the user feature information in the advertisement pool; the user feature information is integrated into user feature vectors, the user feature vectors stored in the edge devices are uploaded to the cloud platform, the cloud platform analyzes the user feature vectors uploaded by the edge devices to generate user portrait information, and the user portrait information is displayed on the basis of the user portrait information generated by the cloud platform. And putting corresponding advertisements to the user groups with similar user portrait information, collecting feedback data, and dynamically optimizing the advertisements put by the edge devices and the user groups based on the feedback data. The method has the advantages that the data transmission delay and the cloud computing pressure are reduced through the cloud edge collaboration method.

The invention relates to a multi-modal large model robot control method based on meta-learning fine tuning, and the method comprises the steps: carrying out the meta-learning pre-training of an OpenVLA model, and obtaining a pre-trained VLA model; performing task specific fine tuning to obtain a fine-tuned VLA model; and inputting the current visual data, the language cue word, the robot sensor data and the task target data graph into the fine-tuned VLA model, and outputting to obtain a decision control instruction for correspondingly controlling the working state of the robot. Compared with the prior art, the method has the advantages that meta-learning pre-training is carried out on diversified tasks, and universal features and strategies are extracted, so that the generalization ability of the model is improved; in addition, in a specific task, rapid fine tuning is performed through a small number of new samples, so that the model can adapt to new task requirements in a short time; and multi-modal information of vision, character cues and robot control is combined, so that the perception and decision-making capabilities of the model are improved, and the adaptability and operation efficiency of the robot in a complex environment are effectively improved.

The invention relates to a knowledge-enhanced large model data analysis agent visualization method, which comprises the following steps of: performing syntactic analysis by using a PCFG algorithm, identifying syntactic components and relationships of sentences, and splitting the sentences into sub-sentences with complete logic; extracting keywords and entity words of the sentences by combining a word segmentation technology with a business word segmentation library; recommending and summarizing candidate segmented words through a recall strategy, carrying out rough arrangement, fine arrangement and fine arrangement on the candidate segmented words, and correcting and complementing a segmented word result; splitting and analyzing the natural language problem based on the fine-tuned large model, and generating SQL statements meeting business requirements; analyzing the data set generated by the SQL, extracting field information and business meanings, and identifying an association relationship between fields; and automatically recommending chart styles according to data features and user preferences, assembling an analysis report and distributing the analysis report according to a specified frequency. According to the method, knowledge enhancement and large model technologies are combined, so that the efficiency and effect of data analysis and visualization are remarkably improved.

The invention provides a visual fault detection method and system for an air conditioning unit, and the method comprises the steps: collecting multi-source data of the air conditioning unit, storing the multi-source data in a matrix form, and then carrying out the preprocessing of the stored data; carrying out feature adaptive fusion to obtain a comprehensive feature vector, and carrying out feature enhancement, dimensionality reduction and regularization processing on the comprehensive feature vector to obtain a final comprehensive feature vector; constructing a causal relationship model, calculating a causal effect, carrying out multi-level fault diagnosis according to a result of the causal effect, and constructing a fault prediction model to carry out fault inference; performing local model training on each edge computing node according to results of the fault diagnosis and fault prediction model; and constructing a fault thermodynamic diagram generation algorithm to perform visualization processing on the fault, designing a multi-mode intelligent interaction interface, and adjusting a visual interface and parameters according to behaviors and feedback of the user. According to the method, multiple innovative technologies such as multi-source data fusion, edge calculation, causal inference and federated learning are introduced, so that the system fault positioning and prevention capabilities are improved.

The invention belongs to the field of switching chips, and relates to a message processing and transmitting device on a CXL switching chip, which comprises a CXL switching interface unit module, an annular topological network on chip and a CXL management unit module, the CXL switching interface unit module is used for receiving and analyzing messages of two sub-protocols, namely. Cache and. Mem, sent by CXL equipment directly connected to a CXL switching chip, then converting and packaging the messages into a message format conforming to the standard of the network on the circular topology chip, and further transmitting the messages through the network on the circular topology chip; and meanwhile, the CXL switching interface unit module also needs to receive the message from the ring topology network-on-chip, converts and packages the message into a corresponding CXL sub-protocol message format, and then distributes the message to other CXL equipment. The method has the characteristics of high bandwidth, low delay and multi-channel concurrency, and provides an efficient transmission channel for a large amount of memory data access scene; and the method has the advantages of high expandability, friendliness to chip back-end physics and the like.

The invention discloses a data cooperative transmission method and system for a low earth orbit satellite Internet of Things terminal, and the method specifically comprises the steps: carrying out the classification cutting of environment data, and forming a plurality of environment data subsets; determining a plurality of cooperative sending terminals and respective optimal communication paths between the first low earth orbit satellite Internet of Things terminal and each cooperative sending terminal based on a communication topological structure relationship between the low earth orbit satellite Internet of Things terminals in the ad hoc network; transmitting the environment data subsets with different priorities to corresponding cooperative sending terminals through different optimal communication paths; and according to the environment data subsets sent by each cooperative sending terminal and received by the low-orbit satellite, splicing the environment data subsets and forwarding the spliced environment data subsets to a target terminal or a ground station or a data center. According to the invention, the cooperative effect among a plurality of low-orbit satellite Internet of Things terminals can be fully utilized, and stable and reliable data transmission under the requirements of large data volume and high real-time performance is realized.

The invention discloses a personalized recommendation system based on semantic analysis, and relates to the technical field of personalized recommendation systems.The personalized recommendation system comprises a data acquisition module, a dynamic situational semantic modeling module and a recommendation engine module, the data acquisition module transmits acquired user behavior data and situational data to the dynamic situational semantic modeling module, and the dynamic situational semantic modeling module transmits the acquired user behavior data and situational data to the recommendation engine module; the dynamic situational semantic modeling module extracts sequence features of user behavior data through a behavior context analyzer, generates context embedding through a situational semantic embedding network, completes cross-modal semantic alignment of multi-modal data through a multi-modal fusion layer, and generates multi-level semantic embedding features. According to the method, the dynamic situational semantic modeling module is designed, the semantic alignment function of the situation and the behavior is achieved, the problem that a traditional recommendation system is insufficient in situation data utilization is solved, the adaptability of the recommendation result to the dynamic situation is enhanced, and the method is suitable for dynamically changing user demand scenes.

The invention relates to the technical field of mixed reality, in particular to a mixed reality multi-user interaction large-space positioning, aligning and collaborating system and method, and aims to solve the problems that in the prior art, the user interaction experience is improved only by collecting multi-aspect data for processing, efficient collaboration work of multiple users in a large space cannot be achieved, and the user experience is poor. Information sharing, task distribution and real-time interaction among users cannot be realized; information sharing, task distribution and real-time interaction among users are realized through the multi-user cooperation module, cross-device operation consistency is ensured through a multi-device sharing coordinate system, configuration is simplified, cooperation efficiency is improved, real-time data transmission is realized through multi-device network synchronization, delay is reduced, cooperation fluency is enhanced, and user experience is improved. The interaction reality sense is enhanced through physical interaction logic synchronization, the working process is optimized, and the cooperation experience is improved.

The invention provides an electric engine, a control method, a related device, a motor controller and an aircraft, and relates to the technical field of aircrafts. The electric engine comprises a power motor, a heat dissipation system and a control module, a cavity is formed in the power motor, the control module is integrated in the cavity, the space occupied by the electric engine is reduced, and therefore the size of the aircraft is reduced; the control module comprises at least three integrated driving modules, part of the driving modules are electrically connected with the power motor so as to control the power motor, part of the driving modules are electrically connected with a variable pitch motor of the propeller so as to control the variable pitch motor, and part of the driving modules are electrically connected with a heat dissipation motor in the heat dissipation system so as to control the heat dissipation motor. The performance and efficiency of the electric engine are improved through unified integrated control over the power motor, the heat dissipation motor and the variable pitch motor. And through a unified control architecture, the system design is simplified, the development and maintenance cost is reduced, and the maintainability of the electric engine is improved.

The invention discloses a communication transmission control cabinet based on big data, and relates to the technical field of communication transmission, and the control cabinet comprises the steps: introducing a real-time network state perception and prediction module, collecting the data of a link through an embedded sensor, carrying out the prediction through combining with an LSTM time sequence model, carrying out the estimation of a future link state, introducing a multipath transmission mechanism, and carrying out the real-time network state perception and prediction. When tasks are distributed, bandwidth, delay and load conditions of paths are synthesized, batch uploading is performed through caching and compression strategies of edge computing nodes, occupation of links is effectively reduced, when link performance is reduced, a path switching mechanism can select alternative paths with optimal performance in real time, influences of link interruption on communication tasks are avoided, and meanwhile, the communication efficiency is improved. According to the high-real-time task priority strategy, an optimal path is adopted, delay of key tasks is reduced, non-real-time tasks are compressed and processed in batches at edge nodes, resources are not competed with real-time tasks, interference is avoided, and the requirements for ship navigation and real-time monitoring tasks are met.

The invention relates to the technical field of industrial internet data acquisition, in particular to a data acquisition method and system based on an industrial internet operating system, and the method comprises the following steps: S1, detecting and recognizing data sources of a plurality of industrial devices and sensors; s2, collecting original data from a data source in real time, and preprocessing the original data; s3, distributing part of calculation tasks to a plurality of edge nodes; s4, storing the data in a distributed database or cloud storage in a classified manner; s5, performing fragmentation processing on the data in the cache; s6, performing data synchronization and consistency verification; and S7, converting the data into a corresponding format. According to the invention, by implementing efficient data fragmentation processing, a dynamic storage strategy and secure data consistency verification, the processing speed and storage efficiency of the industrial internet data acquisition system and the security and integrity of data are remarkably improved, and the requirements of high efficiency and high reliability of modern industrial production are effectively supported.

The invention relates to the technical field of power distribution network recovery, in particular to an urban power distribution network toughness recovery management system under extreme weather, which comprises the steps of acquiring voltage, current, temperature, humidity and wind speed in real time, cleaning, formatting and standardizing data, identifying abnormal conditions influencing the stability of a power grid, evaluating the performance of the power grid under the extreme weather, and managing the toughness of the power distribution network under the extreme weather. An optimization suggestion is provided, various extreme weather scenes are simulated, feedback data are generated, comprehensive risk assessment is carried out, a coping strategy is generated, a fault point needing emergency processing is identified, recovery resources are dispatched, and a result is analyzed in real time to dynamically optimize an operation strategy of a power grid; according to the invention, by introducing the scene simulation and virtual training module, complex dynamic scene simulation is carried out, and coping strategies are verified and optimized. Feedback data generated by the system in simulation is used for further optimizing strategies in the toughness evaluation and decision support module, and adaptability and effectiveness of the strategies in actual extreme weather events are ensured.

The invention provides a ship remote real-time early warning system based on cloud computing, and relates to the technical field of data processing, and the system comprises a data collection module which is used for collecting ship operation data, environment state data and external interaction data; the data preprocessing module is used for generating standard processing data; the data grouping module is used for dividing a priority data group and a common data group; the data transmission module is used for transmitting the data to the cloud analysis module; the cloud analysis module is used for carrying out multi-source data aggregation processing on the priority data set and the common data set, carrying out real-time evaluation on the ship operation state, identifying abnormal behaviors and predicting emergencies, and generating corresponding final early warning instructions including an avoidance operation instruction and an air route adjustment suggestion; the early warning feedback module is used for sending the early warning instruction to the ship terminal; the local response module is used for executing instant triggering operation on the instruction classification data; according to the invention, autonomy and accuracy of remote real-time early warning of the ship are improved.

The invention provides a multi-agent cooperation system, which comprises a multi-agent cooperation engine module, a task decomposition and distribution module and a tool calling normal form module, and is characterized in that the task decomposition and distribution module is used for decomposing tasks and distributing decomposed sub-tasks to agents; the multi-agent cooperation engine module is used for managing cooperation and communication between agents; and the tool calling module enables the intelligent agent to dynamically call different tools. According to the multi-agent collaboration system, the collaboration efficiency among the agents, the flexibility of task decomposition, the adaptivity of tool calling and the intelligent level of user interaction are remarkably improved, and remarkable technical advantages and effects are shown in complex task processing and dynamic environment processing.

The invention belongs to the technical field of copper-clad plates, and relates to a low-dielectric high-heat-resistance copper-clad plate and a preparation method thereof. By introducing the POSS resin, the dielectric constant of the copper-clad plate is effectively reduced, which is crucial for reducing signal delay and improving signal transmission efficiency. The introduction of the POSS resin not only improves the dielectric properties of the material, but also improves the glass transition temperature (Tg) of the material, which means that the physical and chemical properties of the material can be kept unchanged at a higher temperature, and the material is suitable for wider environmental conditions. By using the resin and the inorganic filler in a specific ratio, the copper-clad plate shows excellent copper foil peel strength, and ensures that the bonding force between the copper foil and the base material is not easily weakened under long-term use or severe conditions. Meanwhile, the material also has excellent CAF (conductive anode wire) resistance, can effectively prevent internal short circuit caused by moisture, and prolongs the service life of an electronic product.

The invention discloses a multi-GPU and multi-SSD-based graph neural network training acceleration method and system, and belongs to the field of graph neural network training. According to the method, physical architecture information of a server is read, a communication link relation among a CPU, a GPU and an SSD is modeled into a communication topological graph, the link process that the GPU pulls data is converted into the maximum flow problem of a single source and a single sink, an optimal access flow distribution scheme is solved through Ford-Fulkerson augmented path calculation, and the access flow distribution efficiency is improved. Then, considering the access frequency of the graph vertexes, and distributing the graph vertexes to the storage node with the optimal current performance index for storage; and finally, after storage of all graph vertexes in the graph data set is completed, the server executes batch-by-batch training tasks of the graph neural network. According to the method, the training throughput of the system can be greatly improved, training of TB-level super-large-scale graphs is met, and efficient and low-delay multi-GPU and multi-SSD data access is achieved.

The invention provides a multi-data fusion method and system based on edge computing. Belongs to the technical field of edge computing. The method comprises the following steps: receiving heterogeneous resource data from different sensors, equipment and sources through an edge task allocation unit; according to the encryption level corresponding to each task in the heterogeneous resource data, dividing each task data into data groups with identifiers according to a preset segmentation strategy; according to the operation condition of the edge task processor, performing distribution calculation on the data set according to a preset distribution strategy; after the distribution is completed, obtaining an identification sequence of the distribution mode; the distributed data set is calculated and processed through the edge task processor, and after data calculation and processing are completed, the result in the data set is sent back to the edge task distribution unit. Through intelligent scheduling of the edge task distribution unit, data from different sensors and equipment can be received and processed in real time, so that the data processing efficiency and response speed are improved.

The invention belongs to the field of PCIe Switch interconnection, and relates to a GPU cross-host communication interconnection system based on PCIe NTB, which comprises two hosts, two GPUs and two PCIe Switch chips, the two GPUs are respectively connected with the PCIe Switch chips, the two PCIe Switch chips are connected through a Crosslink line, a connection port of each PCIe Switch chip is set to be in an NTB mode, and the two GPUs achieve interconnection communication through the NTB. By windowing in the NTB, the address of the remote equipment is mapped in the local address space, so that the purpose of accessing the remote equipment is achieved. The PCIe Switch is used for replacing high-end network equipment, so that the hardware cost can be remarkably reduced, and the complexity of system deployment is reduced. Communication between the devices can be directly accessed through a memory address, and complex network connection, IP addresses or special software protocols do not need to be set; and low-delay and high-bandwidth GPU cross-host interconnection communication is achieved. Through the GPU which carries out cross-host communication in a PCIe NTB mode, the effect of GPU Direct P2P can be achieved.

The invention relates to the field of medicine, and discloses a thyroid nodule surgical operation planning and navigation system based on artificial intelligence, which comprises a multi-modal medical image acquisition module, an image preprocessing and enhancing module, an intelligent anatomical structure analysis module, an operation planning decision module and an intraoperative real-time navigation module, the multi-modal medical image acquisition module is used for integrating data of various image equipment such as ultrasonic equipment, CT equipment and MRI equipment, and space-time consistency of multi-modal images is ensured. According to the thyroid nodule surgical operation planning and navigation system based on artificial intelligence, the real-time calculation problem is solved, the system delay is reduced, which is crucial to real-time navigation in an operation, the real-time performance and the operation efficiency are improved, the operation time is shortened, and complications are reduced; by integrating the advantages of different images, the system can provide more comprehensive dissection views, help doctors to more accurately position nodules and important structures and avoid injury to nerves and blood vessels, so that the accuracy is improved, and the surgical risk is further reduced.