305 results about "Collaborative computing" patented technology

The invention discloses a data annotation method and system of a collaborative computing architecture based on quantum computing, and belongs to the field of data annotation. The method comprises the steps that S1, multi-modal data are input and preprocessed; s2, extracting features of each mode after preprocessing; s3, coding the features of each mode into a quantum state, and carrying out mode fusion; s4, performing label reasoning on the quantum state after modal fusion, and performing label constraint optimization by using a quantum approximate optimization algorithm; s5, based on a quantum Bayesian network or an approximate causal graph generation method, generating explanation according to a modal contribution causal path, and deducing marginal contribution of each modal to final label prediction by using a joint probability measurement result; and S6, outputting a labeling result. According to the method, a quantum-classical cooperative computing architecture is designed, the efficiency and accuracy of multi-modal data labeling are remarkably improved, the interpretability, the distributed processing capacity and the high-dimensional feature modeling capacity of the system are enhanced, and a brand new solution thought is provided for development of the multi-modal labeling technology.

The invention discloses a dynamic scheduling method and system for a satellite-ground cooperative computing task, and the method comprises the following steps: constructing a satellite-ground cooperative computing system comprising a remote sensing satellite cluster, a satellite server node, a ground server / data center and the like, and enabling the satellite cluster to generate an in-orbit task and to describe dependence through DAG; available computing power, task queues and other resource states of a satellite and a ground server are collected in real time, and global observation is formed; on the basis of an MADRL framework, each node is regarded as an independent agent, a task allocation action is generated through a strategy network, and a collaborative decision is modeled through POMDP; optimizing the task execution sequence of the same node by using an EFT algorithm; issuing a task and monitoring execution; task completion time and energy consumption are collected, and a network optimization strategy is updated through a reward function and experience playback; the scheduling strategy is adjusted through loop iteration, satellite offline and link fluctuation are adapted, and the resource utilization rate and the task processing efficiency are improved.

The invention relates to an aviation equipment manufacturing digital main line engine system supporting multi-source heterogeneous data access, and belongs to the technical field of aviation manufacturing data management. The system comprises a multi-source heterogeneous data access adaptation module, a data fusion processing module and a full-process data link construction module. The multi-source heterogeneous data access adaptation module passes through a data interface protocol and format conversion component; the data standardization and fusion processing module completes association fusion through an aviation heterogeneous data collaborative computing framework by means of an aviation manufacturing field data element standard library, a unified heterogeneous data model is generated, and the framework comprises a data fragmentation layer, a parallel node layer and a result aggregation layer; and the full-process data link construction module constructs a data link covering the full life cycle of the product based on a time sequence association algorithm and a product unique identifier mapping mechanism. The system can effectively solve the integration and management problems of multi-source heterogeneous data in aviation equipment manufacturing, and improves the data processing efficiency and the whole-process data tracing capability.

The invention discloses a self-adaptive battery life calibration method for cloud-side cooperation, and belongs to the crossing field of an energy storage system and cloud-side cooperation calculation. According to the invention, a cloud-edge double-layer collaborative framework is provided; an edge end estimates the health state and the residual life of a battery in real time through a recursive least square extended Kalman filtering model; the error observer calculates a prediction error based on a sliding window, and a dynamic threshold triggers an uploading mechanism; the edge end adopts an auto-encoder to compress original time sequence features into abstract vectors, and the abstract vectors and error statistics are uploaded together; the cloud performs incremental learning by using a deep sequential network, and only finely adjusts tail level parameters of which the gradient sensitivity exceeds a threshold value to generate a correction value; and the correction value is compressed and issued to an edge end, local model parameters are updated through weighted fusion, and a covariance matrix is adjusted. The method realizes high-precision life prediction and dynamic calibration, remarkably reduces the communication load, and is suitable for electric vehicles, power grid energy storage and other scenes.

The invention discloses a self-adaptive 4D Gaussian splashing high-precision three-dimensional reconstruction system and method, and belongs to the technical field of computer graphic processing. The invention aims to realize high-precision automatic registration of multi-source heterogeneous data and improve the calculation efficiency. The method comprises the following steps: collecting multi-source heterogeneous data; constructing a multi-modal fusion registration method, which comprises the following steps: combining satellite image data and low-altitude oblique photography data to realize spatial distribution geometric coarse registration, fusing low-altitude laser radar point cloud data and ground acquisition vehicle laser radar point cloud data to realize luminosity fine registration, establishing semantic features to assist registration, and obtaining registered multi-source data; initializing a 4D Gaussian primitive and executing adaptive splashing reconstruction to obtain an optimized 4D Gaussian splashing model; designing a cloud edge cooperative computing architecture oriented to 4D Gaussian splash reconstruction, and performing distributed parallel processing on the obtained optimized 4D Gaussian splash model; and executing quality evaluation and adaptive optimization, and outputting a final adaptive 4D Gaussian splash model.

The invention belongs to the field of edge computing acceleration, and relates to a fast Fourier transform and neural network reasoning collaborative computing acceleration method, which comprises the following steps of: deploying a computing method which is based on butterfly computing merging and a tensor mapping strategy and is mixed with DFT (Discrete Fourier Transform) and FFT (Fast Fourier Transform) in an operator deployment level; on the interface integration level, a bus interface of a register access path in a tensor accelerator control path is subjected to lightweight reconstruction, so that the bus interface is adaptive to an edge computing platform; in an operation level, a user-defined instruction is introduced into a tensor calculation unit, and an accelerator is enabled to independently complete a whole-process task of FFT signal processing and NN intelligent identification. According to the method, an operation instruction set is expanded on hardware, and delay overhead caused by returning a large amount of intermediate data to a processor is avoided, so that efficient execution of FFT and neural network tasks on unified hardware is guaranteed, and multiple requirements of high real-time performance, high precision and low power consumption are met.

The invention discloses a battery health management system and method, and belongs to the technical field of battery management, the system comprises an implantable sensor network, an edge and cloud collaborative computing platform, a multi-modal prediction model and a dynamic optimization execution unit; the implantable sensor network integrates various sensors to collect multi-physical field parameters in a battery, and the energy collection unit supplies power through cross-shielding communication transmission. The edge and cloud collaboration platform processes data, trains a model and carries out block chain evidence storage; the multi-modal prediction model is fused with space-time double-flow Transform and causal reasoning, SOH and RUL are predicted, and thermal runaway is early warned; the dynamic optimization execution unit realizes efficient control through layered equalization and self-adaptive thermal management; the method comprises an initialization stage, an operation stage and a maintenance stage to form a closed loop. The method breaks through the traditional limitation, improves the safety, reliability and economy of the battery, and is suitable for electric vehicles, energy storage power stations and other scenes.

The invention relates to the technical field of cloud edge collaborative computing power scheduling, and discloses a heterogeneous computing power scheduling optimization method based on a cloud edge collaborative architecture. The method comprises the following steps: acquiring real-time computing power state data of all available computing nodes in the cloud edge collaborative architecture; performing heterogeneous type division on the computing nodes according to the real-time computing power state data to generate a three-layer computing power resource pool containing cloud computing nodes, edge computing nodes and terminal computing nodes; extracting task calculation features for the current to-be-scheduled task set, wherein the features comprise calculation intensity, data dependence and real-time requirements; constructing an initial task allocation scheme based on the matching relationship between the task calculation features and the three-layer computing power resource pool; iteratively optimizing the initial scheme by adopting a dynamic load balancing strategy to generate a final task scheduling instruction; the instructions are distributed to the corresponding computing nodes to be executed, and computing power state changes in the task execution process are continuously monitored.

The invention relates to the technical field of monitoring, and discloses a multi-modal video public opinion intelligent monitoring method and system, and the system comprises a multi-modal data collection module, a data processing module, a coefficient generation module, an intelligent analysis module, a dynamic threshold module, and a public opinion visual platform. According to the method, a visual, voice and character cross-modal cooperative computing framework is established, and the one-sidedness problem of traditional single-modal analysis is solved; a recognition-tendency-propagation three-dimensional index system is constructed, and accurate quantitative evaluation of public opinion evolution is achieved; designing a static comprehensive coefficient threshold value and a dynamic growth threshold value of the dual-threshold grading system, and cooperating with an imaginary main body filtering algorithm and KOL polarization value calculation, so that the false alarm rate is remarkably reduced, and meanwhile, rapid early warning response is realized; a three-dimensional propagation map is generated through propagation path concentration analysis, content feature evolution tracking and influence radiation, and a complete public opinion evolution chain and hotspot traceability are provided.

The invention discloses an intelligent video monitoring system and method supporting AI strategy cloud side end cooperative execution. According to the system, unified management of computing power resources and dynamic deployment of an AI algorithm are realized based on an extended standardized protocol through a three-level architecture of a cloud side center platform, an edge computing node and an end side camera. The method comprises the following steps: a cloud platform acquires computing power information of edge equipment; an AI algorithm execution strategy is generated and issued according to service requirements; if the required algorithm is not deployed in the equipment, the equipment is controlled to download and automatically deploy an algorithm module; and finally, the driving equipment executes AI analysis and recovers the structured data. According to the invention, AI computing tasks are reasonably distributed through a cloud side-end cooperative computing architecture, so that the computing power construction cost of a central platform is effectively reduced; through algorithm and hardware decoupling and a standardized protocol interface, ecological binding is broken, on-demand loading and flexible scheduling of the algorithm are realized, and the economical efficiency, the flexibility and the intelligent level of the system in large-scale scenes such as smart cities and the like are remarkably improved.

The invention discloses an SM2 collaborative signature and encryption and decryption system and method fusing anti-quantum characteristics, and relates to the field of cryptography and information security. According to the method, an anti-quantum cryptographic algorithm and a national cryptographic SM2 cooperative computing framework are deeply integrated, and a key security system is constructed: SM2 sub-private keys, anti-quantum key pairs and public keys are acquired and generated through an anti-quantum algorithm software and hardware enhancement module, and the private keys are encrypted and stored and are regularly alternated; on the basis of anti-quantum collaborative signature, encryption and decryption modules, an anti-quantum verification mechanism is embedded, and data is transmitted in combination with a national secret TLCP protocol, so that signature, encryption and decryption operations are completed; the equipment integration and dynamic security control unit monitors a security state, calculates a threat index to generate a protection strategy, and dynamically switches a security mode, so that the problems of insufficient security, vulnerability to attacks and data tampering of a traditional SM2 algorithm under the threat of quantum computing are effectively solved; and the long-term anti-attack capability and the operation reliability of the equipment in a high-security demand scene are remarkably improved.

The invention discloses an intelligent transport capacity dispatching system and method based on a GPS. The system comprises a transport capacity state monitoring subsystem and an intelligent dispatching management subsystem. And by deploying a GPS positioning and multi-parameter sensor, real-time acquisition and analysis of information such as the position, the speed and the energy consumption of the transport capacity unit are realized. According to the system, CNN and Transform structure extraction features are fused, LSTM is combined to predict a transport capacity demand, and a PPO reinforcement learning algorithm is adopted to dynamically plan a path. Abnormality detection (SVM), PID feedback adjustment and cloud edge cooperative calculation are integrated in the scheduling process, and the scheduling precision and the response speed are improved. The system has the capabilities of task filing, online optimization and data security guarantee, and is suitable for complex traffic and logistics scheduling scenes.

The invention discloses a cloud side-end collaborative architecture and task unloading method for an airport apron intelligent monitoring system, and relates to the technical field of cloud side-end collaborative computing. Comprising a terminal layer, an edge server layer and a cloud server layer, the task unloading method of the system under the cloud side-end collaborative architecture is designed and comprises the steps that a task model, a time delay model, an energy consumption model and an accuracy rate model of the system are constructed, and a multi-objective optimization problem of time delay-energy consumption-accuracy rate is formed; modeling an optimization problem into a Markov decision process, and designing a state space, an action space and a reward function required by deep reinforcement learning; and designing a task unloading method based on multi-agent deep reinforcement learning, and finding an optimal task unloading strategy of the system. According to the method, the computing tasks can be dynamically, scientifically and reasonably distributed and cooperatively scheduled among the terminal, the edge and the cloud, so that the time delay, the energy consumption and the accuracy are comprehensively optimized, and the overall performance of the system is improved.

The invention discloses a property asset digital operation and intelligent security linkage cooperative processing system and an implementation method, and relates to the technical field of Internet of Things, and the system comprises a protocol adaptive gateway layer which solves the problem of compatibility of multi-manufacturer equipment, and outputs a standardized data stream based on a dynamic protocol sniffing and fingerprint feature library, SDK-free access control, fire protection, monitoring and IoT equipment; the method comprises the following steps: constructing a millimeter-level precision three-dimensional space model by a digital twinborn middle table layer, fusing a security event and operation data, and ensuring that the operation is compliant and audible through blood relationship tracking; the intelligent decision execution layer combines a rule engine and machine learning to generate a linkage instruction, drives a work order to be automatically distributed and positions an equipment fault responsible party; and the edge cooperative computing node locally realizes video millisecond-level analysis, and starts localized linkage according to an emergency plan when the network is disconnected, so as to ensure that the core security function is not interrupted.

The invention relates to the technical field of natural language processing, and discloses a special disease knowledge spectrogram construction and analysis system, which comprises a federal data acquisition module, a dynamic ontology mapping preprocessing module, a cross-source knowledge collaborative extraction module, a self-adaptive graph construction module, an intelligent dynamic updating module and an interactive analysis application module, the cross-source knowledge collaborative extraction module is used for collaborating multi-source data on the premise of guaranteeing data privacy, the dynamic ontology mapping preprocessing module is used for mapping knowledge representation of the multi-source data to a unified standard, and the cross-source knowledge collaborative extraction module is used for extracting medical entities and semantic relationships based on multi-modal data fusion. According to the method, through the cooperation of the longitudinal federated learning framework and the secure multi-party computing algorithm, the problem of multi-source medical data privacy collaborative computing is solved, and the advantages of safely aggregating multi-mechanism data and providing original data support for map construction on the premise of ensuring data privacy are achieved.

According to the decentralized AI model optimization method based on the block chain, the block chain is used as an infrastructure, chain evidence storage of training task parameters, node computing power and training logs is realized based on an intelligent contract, and the whole training process is ensured to be traceable and tamper-proof; dynamic separation and distribution of the sub-models are realized, and the resource utilization efficiency in a heterogeneous environment is improved; constructing a decentralized pipeline training mechanism, supporting cross-node cooperative computing and intermediate activation transmission, and guaranteeing the transparency and credibility of the training process; model watermark dynamic detection, training log hash evidence storage and delay verification mechanisms are fused, a multi-level training behavior supervision system is constructed, abnormal or malicious nodes are effectively recognized and restrained, and fairness, safety and sustainable development of an open AI training platform are promoted.

The invention discloses an electric ship remote intelligent control method and system based on cloud-side cooperation, and relates to the technical field of cloud-side cooperation computing, and the method comprises the steps: carrying out the feature extraction of images around a ship and time-space aligned multi-modal data, obtaining visual features, motion features and environment features, carrying out the feature fusion, and generating an environment situation vector; inputting the environment situation vector into a channel boundary detection model, analyzing the emergency degree of the current situation, generating a collision avoidance instruction, and generating a ship control decision through a three-level collaborative decision; combining a ship control decision with ship control characteristics, simulating a traffic development situation in a navigation area, comparing with historical ship navigation conflict data, and identifying potential conflict data; through collaborative fusion of the ship-end decision-making unit, the shore-end decision-making unit and the cloud-end decision-making unit, an optimal control strategy is generated in combination with respective advantages, intelligent decision-making of global optimization in a complex environment is realized, and adaptability and safety of ship control are enhanced.

The invention discloses a high-speed target continuous tracking system and method based on inter-satellite cooperative computing. The system comprises a distributed architecture management module, an inter-satellite communication module, a double-satellite cooperative tracking module and a distributed computing coordination module. The system provided by the invention adopts a distributed hierarchical management structure, and through a hierarchical management mechanism of intra-cluster local management and inter-cluster global coordination, not only are the quick response and autonomy of a distributed system reserved, but also the global management capability of a centralized layer is introduced, and the communication overhead is effectively reduced. In addition, a multi-type missile trajectory prediction method based on deep learning is designed, a prediction-driven tracking switching mechanism is achieved, the next round of prediction task is triggered through an adaptive threshold value, and missing of a visibility window is avoided. In the aspect of computing resource scheduling, the time delay of task scheduling can be reduced, the communication overhead is remarkably reduced while the tracking precision is ensured, the high timeliness of the tracked target is ensured, and the global continuous tracking of the target is realized.

The invention relates to the technical field of electric power artificial intelligence, in particular to an electric power artificial intelligence-oriented server and board card collaborative computing power scheduling method and system, and the method comprises the steps: receiving an electric power AI computing task, and carrying out the analysis of the task features of the task; monitoring the real-time resource state of the server system in real time; based on the task features and the real-time resource state, generating a collaborative scheduling strategy through an intelligent scheduling decision engine; generating and issuing a corresponding collaborative configuration instruction set according to the collaborative scheduling strategy; and the target AI intelligent computing board card executes a first action according to the computing power configuration instruction, and meanwhile, the server CPU executes a second action coordinated with the first action according to the coordinated resource instruction so as to jointly complete an electric power AI computing task. According to the technical scheme, the collaborative breakpoint problem in a task execution link is effectively solved, and refined collaborative scheduling of the server CPU and the AI intelligent calculation board card is achieved.

The invention provides a field section turnout intelligent control and informatization management method and system, and relates to the technical field of computers, and the method comprises the steps: obtaining a basic data set in a target field section; performing data fusion according to the basic data set to obtain equipment real-time state information; performing health diagnosis according to the real-time state information of the equipment to obtain an equipment health degree evaluation result; intelligent decision making is carried out according to the equipment health degree evaluation result, and a self-adaptive control decision instruction is obtained; performing digital twinborn verification according to the adaptive control decision instruction to obtain an optimized safety control instruction; edge execution is carried out according to the optimized safety control instruction, the instruction is transmitted to a local node through a cooperative computing architecture, and the closed-loop control state of field section operation is obtained. According to the method, the whole-process integrated intelligent management and control of field section operation from state perception, intelligent diagnosis, active decision making to safe execution is realized, and the reliability, the safety and the operation and maintenance efficiency of the system are remarkably improved.

The invention relates to the technical field of integrated circuits and distributed computing, in particular to a chip computing network-oriented clock synchronization method, which is convenient for a receiving end to quickly identify and reduce analysis overhead by customizing a simplified synchronization frame format. A hardware synchronization module is deployed between a physical layer and a link layer to record a high-precision timestamp to eliminate an error caused by protocol stack processing, and a slave node calculates clock skew based on timestamp information interacted by a master node and a slave node and corrects the clock skew in real time to realize accurate synchronization. Master-slave roles are independently configured through each communication port, and automatic topology discovery is supported to adapt to a complex network structure. According to the method, an existing routing protocol and a physical link of the chip are reused, hardware overhead is reduced, synchronization precision is high, flexibility is high, the method can be widely applied to a computing network formed by an artificial intelligence acceleration chip graphics processor, a data processor and the like, and multi-chip cooperative computing efficiency and data consistency are effectively improved.

The invention discloses a computing power network hierarchical scheduling optimization method and system based on cloud side end collaboration, and a medium, and the method comprises the steps: S01, obtaining the information of nodes in a computing power network, the computing power network employing a cloud side end three-layer computing architecture; step S02, establishing a DAG according to the dependency relationship among the task nodes; step S03, constructing an MDP model, defining a state space as a combination of DAG information and a task unloading plan, defining an action space as to unload tasks to any layer, defining a reward function as to maximize the service quality of the scheme as a target, and calculating the service quality according to time delay and energy consumption; and step S04, generating a task unloading decision scheme by using the deep reinforcement learning model, and cooperatively completing execution of all tasks according to the task unloading decision scheme. The three-layer computing architecture can be integrated to improve the execution efficiency of the computing task and the resource utilization rate, and the overall performance and flexibility of the system are remarkably improved.

The invention belongs to the field of end-cloud collaborative computing and lightweight virtualization, and provides an end-cloud collaborative lightweight application distribution and isolated operation method and system, which adopts a WASM byte code as a unique format for application distribution, compiles an application into a WASM module at a cloud and then distributes the WASM module to an end side device; the end side equipment combines read-only configuration data of a host with a temporary file system Tmpfs of a memory by utilizing OverlayFS on a file level, and maps the data into a virtual file system in a WASM sandbox; on the symbol level, a dynamic WASI bridge is introduced, according to the requirement of an import section in a WASM module, the native capacity of a host system is injected into WASM operation in a host function mode in real time, and the heterogeneous hardware characteristics of end-side equipment are adapted. The method is suitable for an end-side environment with an immutable operating system as a base, and is compatible with a cloud containerization deployment scene.

The present application relates to a kind of multi-dimensional quality of service driven FSO / RF space-ground integration robust resource allocation method, belong to wireless communication network technical field, the existing FSO / RF SAGIN cloud edge collaborative computing method cannot consider the efficiency of computing task processing, timeliness and computing task transmission reliability and the problem of needing to rely on accurate CSI condition to carry out computing task processing is solved.This application first establishes FSO / RF SAGIN cloud edge collaborative computing network model, and constructs optimization problem OP1 based on the model, to realize the maximum long-time computing task processing amount of all users as the goal, solves optimization problem OP1 using DDPG deep learning method, and the optimal solution is obtained, i.e.FSO / RF space-ground integration robust resource allocation scheme.The present application can ensure the realization of high quality of service FSO / RF space-ground integration cloud edge collaborative computing from multiple dimensions under uncertain channel conditions.

The invention discloses a cloud edge end collaborative computing power scheduling system supporting task dynamic migration, which relates to the technical field of intelligent computing power scheduling and comprises a path modeling module, a conflict identification module, a time sequence checking module, a rehearsal switching module and an interference suppression module, and constructing a routing convergence time sequence profile and a link playback spectrum, modeling an overlapping window of the old path and the new path and a life cycle of a control signal, and generating a path evolution baseline for subsequent conflict identification. By means of path modeling, fingerprint identification, causal check, shadow rehearsal, interference suppression and the like, time sequence consistency of a control instruction and a data stream, continuous path switching and real-time extinguishing of link interference are realized, and the stability and reliability of cloud edge collaborative scheduling are improved.

The invention relates to the technical field of underground water control and intelligent construction of deep foundation pit engineering, and discloses an underground water recharge control system and method. Comprising a water level real-time monitoring module, a central intelligent control module, a recharge well group regulation and control module, a flow dynamic regulation module, a pressure balance compensation module, a servo support linkage module, a risk early warning module and a data visualization module. A cloud-edge cooperative computing architecture is established, intelligent control is realized, and the system has the advantages that firstly, an artificial intelligence algorithm is introduced into the field of hydrogeological regulation and control, and self-adaptive optimization of a recharge strategy is realized; secondly, a coupling control mechanism of recharge pressure and supporting axial force is established, and foundation pit deformation is restrained from the source; and thirdly, through deep fusion of a robot system and a recharge system, a full-process digital construction system is constructed, so that the efficiency, precision and safety of super high-rise building construction are greatly improved.

The invention relates to the crossing field of fault prediction and health management technologies, and discloses a bridge structure health real-time monitoring method and system. The method comprises the following steps: constructing a high-fidelity parameterized finite element model; deploying a multi-source heterogeneous sensor network; constructing a neural network fused with physical constraints; correcting model parameters on line based on a joint loss function; and outputting a health assessment result. The system comprises a sensor network, an edge-cloud cooperative computing architecture, a historical data backtracking calibration module and a multi-scale verification mechanism. By adopting the technical scheme, the self-evolution and high-fidelity maintenance of the digital twin model can be realized, and the real-time performance, the reliability and the engineering practicability of bridge health monitoring are improved.

The invention discloses a satellite network topology discovery method and a path construction method and device, and the method comprises the steps: dividing a satellite network into autonomous domains through distributing an extended SRv6 SID which comprises a network prefix, an orbital plane number, a node number, a service type and a resource margin for each satellite, and building an SID matrix and a satellite node adjacency matrix; the satellite nodes sense own resource states, periodically detect link states by adopting a BFD (Bidirectional Forwarding Detection) protocol, and synchronize values of elements in a matrix through an incremental updating mechanism to realize dynamic topology sensing; meanwhile, a routing path construction method based on an SID matrix and an adjacent matrix is provided, and satellite nodes are supported to carry out path calculation locally. The method supports heterogeneous service identification and resource state awareness, can effectively reduce signaling overhead, improves accuracy and routing efficiency of satellite network topology discovery, and improves cross-constellation computing power pooling and cooperative computing capability.

The invention relates to the technical field of industrial Internet of Things and edge computing, and discloses an edge-end cooperative computing task unloading method and system for an intelligent factory, and the method comprises the steps: building a task unloading decision mechanism of a multi-device multi-edge server, distributing a computing node for each task, determining task delay, and constructing a node speciality matrix; establishing an optimization model; formalizing an optimization problem of the optimization model into a Markov decision process; solving a Markov decision process by using a deep reinforcement learning algorithm, and realizing intelligent workshop environment interaction so as to store generated empirical data into an empirical cache region; synchronously updating network parameters of the deep reinforcement learning algorithm and a node speciality matrix corresponding to the edge server by using the data pairs of the experience cache region; and outputting a strategy network and a node speciality matrix until the deep reinforcement learning algorithm converges. And an optimal task allocation decision is dynamically generated, so that the task completion deviation is effectively reduced.

The invention discloses a dynamic self-adaptive edge-cloud cooperative computing segmentation point and compression strategy adjustment method. The core of the method is to deploy a dynamic decision engine on edge equipment. The engine senses a network state (such as bandwidth and delay) and a task priority specified by an application in real time, and according to a preset decision logic (such as a look-up table or a lightweight prediction model), a deep learning model which supports multiple segmentation points and is equipped with a variable bit rate hierarchical quantization feature compression model for each segmentation point is used for performing a multi-segmentation-point hierarchical quantization feature compression model on the basis of the multi-segmentation-point hierarchical quantization feature compression model, so that the multi-segmentation-point hierarchical quantization feature compression model is obtained. And dynamically selecting an optimal model segmentation point and feature compression level combination. The combination is transmitted to the cloud, and the cloud loads the corresponding back-end model to complete calculation. Through the dynamic adjustment mechanism, the real-time optimal balance among the reasoning delay, the task precision and the network overhead is realized, and the performance stability, the resource utilization rate and the adaptability of the system in a variable environment are remarkably improved.