98 results about "Cloud scheduling" patented technology

The invention provides a multi-robot collaborative scheduling method, system and device for operation and maintenance tasks of a computing power center and a medium, and belongs to the technical field of computing power centers. A cloud scheduling center receives and classifies cloud task requests, generates standardized task description and sends the standardized task description to edge nodes; the edge node realizes local robot task allocation based on a multi-dimensional bidding rule through a local auction device, and generates a preliminary allocation contract; the cloud scheduling center periodically receives the contract abstract, performs cross-node optimization through a global corrector and issues a correction instruction; when the communication is interrupted, the edge node starts the network disconnection self-healing module, and the execution capability is maintained through the local cache, the offline SLAM and the offline task distribution. The operation and maintenance task response speed and the system reliability are improved by combining queue jumping, queue pre-jumping tasks and a network disconnection self-healing mechanism; full coverage of complex operation and maintenance scenes is supported; the offline technology ensures continuous execution of tasks when communication is interrupted, and ensures the operation and maintenance stability of the computing power center.

The invention discloses a cross-cloud task arrangement optimization method based on multi-cloud resource scheduling, and the method comprises the steps: constructing a task topological weighted graph, and recognizing a parallelizable node set; constructing a task node preliminary scheduling feasibility table according to task topological weighted hierarchy information; constructing a granularity feature vector, and updating the task structure matrix; identifying granularity collaborative attenuation abnormal nodes, and performing cross-cloud task path repair and structure update; generating a cross-cloud scheduling time matrix based on each cloud platform resource supply capability index; predicting resource competition conflicts in combination with task execution index historical data, and generating a conflict prediction table; identifying granularity strategy deviation, and determining a repair trigger point; and generating a cross-cloud task granularity adaptive arrangement configuration file based on the repair trigger point information. According to the method, high reliability, high flexibility and high adaptability of task scheduling can be realized in a complex multi-cloud heterogeneous environment, and the stability, the resource utilization rate and the overall cooperation efficiency of cross-cloud task execution are remarkably improved.

The invention relates to the technical field of industrial vehicle intelligence, in particular to an intelligent flexible forklift system based on multi-mode sensing and self-adaptive path planning. The intelligent forklift comprises an intelligent forklift body, a vehicle-mounted integrated control system, a multi-mode sensing and navigation module, a self-adaptive decision-making and path planning module, a cooperative communication and task management module and a cloud scheduling and management platform, and the intelligent forklift body is provided with a pallet fork mechanism with a 3D camera and a force sensor; the vehicle-mounted integrated control system serves as a core to achieve multi-module data interaction and instruction output, the multi-mode sensing and navigation module generates an environment situation map through multi-source data fusion, the self-adaptive decision-making and path planning module completes path planning, obstacle avoidance and pre-verification, the cooperative communication and task management module achieves multi-vehicle cooperation and task disassembly, and the multi-mode sensing and navigation module generates an environment situation map through multi-source data fusion. And the cloud scheduling and management platform realizes global monitoring and management. According to the method, the adaptability to complex scenes and the operation safety are improved, and accurate forking and flexible path planning are realized.

The invention relates to the technical field of industrial network control, in particular to a side-cloud collaborative software-defined industrial control system, which comprises a side control action configuration module, a multi-source signal collection module, a state mutation recognition module, a cloud scheduling module and a collaborative control generation module. In the invention, through decoupling control action configuration, multi-channel signal acquisition, data label construction, sudden change state identification and a remote scheduling execution process and realizing standardized processing, an edge control basic instruction structure body can be quickly constructed based on an equipment state and a unified time sequence data label mechanism is established; in combination with the dynamic sensing ability of a sudden change state and a feedback point identification mode, efficient discrimination and rapid classification of key control points can be realized, task chain execution can be rapidly mapped to a key path and real-time regulation and control are implemented, rapid response to industrial field multi-source change and accurate issuing of a control instruction are realized, and the task chain execution efficiency is improved. And the real-time performance and the collaboration of the control process are greatly improved.

This invention relates to the field of robot scheduling technology, and more particularly to a multi-robot cloud scheduling and collaborative method and system. The method includes the following steps: acquiring building elevator group data, and dividing the building elevator service groups based on the building elevator group data to obtain building elevator service group data; acquiring robot target path data; segmenting the robot target path data and building elevator service group data into path resource segments to obtain split path data; acquiring current transfer floor data, and estimating transfer waiting time based on the current transfer floor data and split path data to obtain transfer waiting time estimation data; constructing a transfer map based on the transfer waiting time estimation data to obtain transfer map data; obtaining path conflict data based on the transfer map data; and obtaining robot scheduling data based on the path conflict data to perform multi-robot collaborative scheduling operations. This invention can improve task completion efficiency and scheduling intelligence in multi-robot cross-floor collaborative scenarios.

The invention relates to the technical field of photovoltaic power generation cloud platforms, and discloses a multi-site energy storage cluster cloud scheduling platform based on photovoltaic power generation prediction, and the platform employs a real-time data collection and space-time synchronization module to carry out the timestamp alignment and space grid processing of multi-source heterogeneous data. In combination with a dynamic priority scheduling mechanism of a prediction task distribution module based on volatility and a power grid security situation, the effects of cooperative execution of prediction tasks and priority guarantee of key tasks are achieved; through containerized parallel reasoning and consistency Hash load balancing technology of a distributed prediction calculation module, time consistency of a multi-node prediction result is ensured, and a rolling optimization algorithm and a time sequence constraint embedding strategy of a unified scheduling decision module are adopted. Accurate calculation of power generation unbalanced power and time sequence consistency generation of an energy storage strategy are achieved, and through instruction verification and conflict detection and filtering frequency reduction processing of a smoothing module, the effects of prolonging the service life of equipment and guaranteeing stable operation of a power grid are achieved.

Disclosed in embodiments of the present description are a data processing method, apparatus and device based on an end-edge-cloud scheduling system. The method comprises: acquiring a first service processing request in the process of executing a target service; then, determining a computing power allocation strategy for a terminal side and an edge server side and corresponding to the first service processing request, wherein the computing power allocation strategy is determined on the basis of the number of service processing requests processed by the terminal side, the number of service processing requests processed by the edge server side, and the trusted release rates of service processing requests corresponding to the terminal side and the edge server side, under the computing power constraint conditions and / or service time consumption conditions of the edge server side; and finally, on the basis of the computing power allocation strategy, determining a device side for processing the first service processing request, and processing the first service processing request on the basis of the determined device side.

The invention relates to the technical field of computers, and discloses a data backup method, device and equipment and a medium, which are applied to a data backup system, and comprise a data label generation module, a cloud resource analysis module, a cross-cloud scheduling module and a data backup module. Generating structured label data based on the key information of the to-be-backed-up data through a data label generation module; determining a target access frequency and a target confidentiality level of the structured tag data through a cloud resource analysis module, and determining path information; generating backup summary information based on the path information and the to-be-backed-up data through the cross-cloud scheduling module, and sending the backup summary information to the data backup module; and carrying out structured storage on the backup summary information through a data backup module, and storing the backup summary information to a distributed block chain so as to back up the to-be-backed-up data. The method can be applied to business program systems such as financial science and technology, medical health care and the like, and the efficiency, accuracy and compliance of data backup can be improved.

The invention provides an ARD printing task processing method and system based on cloud scheduling and a storage medium, and the method comprises the steps: firstly, generating printing task information through analysis of a production order, and dynamically generating an optimal printer sequence through a task distribution network model in combination with an ARD printer cluster state collected in real time; then, the task is converted into a device-level instruction through an edge gateway, a selected printer is driven to execute printing, and a device feedback signal is monitored in real time; in addition, when abnormal conditions such as continuous label taking failure or task timeout are detected, model rescheduling is automatically triggered, a printer sequence is dynamically switched, and tasks are distributed again; according to the invention, the idle time of equipment is reduced through cloud intelligent distribution, the utilization rate of a printer cluster and the response speed of an emergency order are improved, and through a three-layer architecture of cloud decision, edge execution and abnormal closed loop, the pain points of response delay and slow fault recovery of a traditional printing system are solved, and industrial-grade high-reliability printing task scheduling is realized.

The invention relates to the technical field of robot scheduling, in particular to a multi-robot cloud scheduling cooperation method and system. The method comprises the following steps that building elevator group data are obtained, building elevator service groups are divided according to the building elevator group data, and building elevator service group data are obtained; acquiring target path data of the robot; performing path resource section cutting on the robot target path data and the building elevator service group data to obtain split path data; acquiring current transit floor data, and performing transit waiting estimation according to the current transit floor data and the split path data to obtain transit waiting estimation data; carrying out transfer graph construction according to the transfer waiting estimation data to obtain transfer graph data; obtaining path conflict data according to the transfer graph data; and obtaining robot scheduling data according to the path conflict data so as to carry out multi-robot cooperative scheduling operation. According to the method, the task completion efficiency and the scheduling intelligence level can be improved in a multi-robot cross-floor collaborative scene.

The invention relates to a charging pile intelligent distribution system and method, and the system comprises a user terminal module, the output end of the user terminal module is electrically connected with a cloud scheduling platform, and the output end of the cloud scheduling platform is electrically connected with a charging pile control module. According to the charging pile intelligent distribution system, through intelligent dynamic scheduling and elastic time window management, the utilization efficiency of charging resources is remarkably improved, the user waiting time is shortened, the power grid load balance is optimized, the system adopts a multi-target optimization algorithm, the user distance, the real-time electricity price and the charging pile load rate are comprehensively considered, and the system reliability is improved. According to the method, fairness distribution is ensured, emergency demand priority response is supported at the same time, scheduling transparency and data security are enhanced through application of the block chain technology, a credit rating mechanism effectively restrains user performance behaviors, and the system can dynamically adjust power distribution and quickly deal with equipment abnormity through cooperation of load balancing and fault self-healing functions, so that the system performance is improved. And the service continuity is ensured.

A method of creating confidential cloud computing platform including at least one of plurality of confidential virtual machines (VMs), in a disaggregated resource architecture. The confidential cloud computing platform receives security policy from user at attestation and key management system that includes public key. The cloud resource gateway receives request to create the confidential VM from user according to the received security policy. The cloud resource gateway request the cloud scheduler system to create at least one confidential VM on top of disaggregated resource architecture and interacts with monitoring agents in a plurality of compute pools comprising processor hardware resources and storage pools comprising storage hardware resources to obtain availability information regarding processor and storage hardware resources, cloud scheduler system allocates processor hardware resources and storage hardware resources to create at least one confidential VM and attestation and key management system performs attestation for allocated processor and storage hardware resources.

The application relates to a cloud service-based model inference service system and method capable of on-demand scaling. The system comprises a gateway module, a model inference service management module and a request distribution module. The gateway module is used for receiving a user request and extracting heterogeneous features representing the request's demand for computing resources. The model inference service management module is used for generating resource scheduling instructions for each inference service instance based on the multi-dimensional indicators of the deployed inference service instances, in combination with the resource cost and availability information of each cloud resource pool. The request distribution module is used for distributing the user request to the inference service instance that is adapted to the heterogeneous features according to the resource scheduling instructions. The system can solve the problems of low utilization rate of computing resources, poor request adaptability and lack of cross-cloud scheduling in the prior art.

This invention discloses a non-embedded UAV information acquisition device and method. The non-embedded UAV information acquisition device includes a shell assembly, a three-mode adapter structure, a standardized physical interface, a main control module, a communication module, a positioning and sensing module, and a power management module. The internal cavity wall of the shell assembly is fitted with an anti-magnetic shielding layer made of permalloy. The three-mode adapter structure is integrated into the bottom of the shell assembly and includes a magnetic suction component, a suspension component, and a screw fixing interface. The standardized physical interface is located on the surface of the shell assembly and uses a gold-plated gold finger contact structure, which is electrically connected to the main control module. This invention enables deployment without disassembly of the UAV through a non-intrusive design while maintaining the UAV's airworthiness. It utilizes edge computing to alleviate cloud scheduling pressure and reduce task latency, and with independent power supply, achieves long endurance and high reliability, making it widely adaptable to various UAV models and harsh operating environments.

The invention provides a UV printer intelligent control system and method based on remote task scheduling, and the system comprises a cloud scheduling center module which is used for global task planning and distribution, network state monitoring, equipment state monitoring, and data storage and analysis; the edge gateway node module is used for caching the task data issued by the cloud scheduling center module, serving as a local scheduling center when the network is interrupted, and performing cooperative control and fault diagnosis and early warning on a plurality of devices at the same time; and the UV printer terminal module is used for collecting state data of printing equipment in real time, receiving and executing tasks, locally caching the data and performing man-machine interaction. By combining cloud global optimization, edge local collaboration and terminal offline autonomy, the dependence on a single network is reduced, the overall production efficiency is improved through multi-device collaboration and load balancing, ink waste and invalid loss of devices are reduced based on a prediction model, a printing path is optimized, and energy consumption is reduced.

The invention discloses a warehouse logistics scheduling method and system based on cloud computing, and relates to the field of warehouse logistics management, and the method comprises the steps: collecting to-be-processed cargo information and available robot state information in a warehouse management system in real time based on a cloud scheduling platform; carrying out the clustering of the to-be-processed goods, carrying out the calculation and evaluation of a clustering result, and obtaining an optimal task partition division scheme; obtaining the shortest collision-free path from each available robot to each task partition, and calculating the path length of the shortest collision-free path; calculating path suitability, load suitability and electric quantity sufficiency between each available robot and each task partition; determining an optimal allocation scheme between the available robots and the task partitions based on the three features; and scheduling the available robots based on the optimal allocation scheme and monitoring the operation states of the robots in real time. According to the invention, real-time, efficient and balanced global integrated scheduling can be carried out on logistics robots and order tasks, and the collaborative operation capability of a robot cluster is improved.

The present application relates to the technical field of computer, specifically relates to a kind of dynamic cloud scheduling method and system of artificial intelligence workflow;The method of the present application includes defining standardized AI-Workflow IR intermediate representation, convert native workflow into the workflow of standardized AI-Workflow IR intermediate representation, build calculation dependency graph, obtain the key path node in calculation dependency graph by CPA algorithm, form key path, mark the subgraph in key path, carry out subgraph segmentation and subgraph optimization to the workflow of standardized AI-Workflow IR intermediate representation using GRR algorithm, based on deep reinforcement learning algorithm, the matching scheduling of Agent is carried out to subgraph;The present application supports the conversion of workflow of different frameworks, optimizes scheduling strategy with environmental change, JIT-Agent generates specialized service on demand, and realizes the distribution of workflow reasonably.

The invention relates to a model reasoning service system and method for on-demand capacity expansion and contraction based on cloud service. The system comprises a gateway module, a model reasoning service management module and a request distribution module, and the gateway module is used for receiving user requests and extracting heterogeneous features representing the requirements of the requests for computing resources; the model reasoning service management module is used for generating a resource scheduling instruction for each reasoning service instance based on the multi-dimensional index of the deployed reasoning service instance in combination with the resource cost and availability information of each cloud resource pool; and the request distribution module is used for distributing a user request to the inference service instance matched with the heterogeneous characteristics according to the resource scheduling instruction. By adopting the system, the problems of low computing resource utilization rate, poor request adaptability and cross-cloud scheduling deficiency in the prior art can be solved.

The invention provides an intelligent calculation center cloud scheduling system, and relates to the technical field of artificial intelligence and big data, and the system comprises a GPU cluster which is formed by the mixed deployment of at least a first type of GPU cards and a second type of GPU cards; and the scheduler is connected with the GPU cluster, a plurality of scheduling strategies are configured in the scheduler, and the scheduler is used for carrying out hybrid scheduling on the first type of GPU cards and the second type of GPU cards based on the scheduling strategies. The method has the beneficial effects that multi-strategy hybrid scheduling of different types of GPU cards is effectively realized, the use efficiency of computing power resources is improved, the operation cost is reduced, the operation efficiency of a system is improved, and the system risk is reduced.

The invention relates to the technical field of computers, in particular to a dynamic cloud scheduling method and system for an artificial intelligence workflow. The method comprises the steps of defining a standardized AI-Workflow IR intermediate representation, converting an original workflow into a workflow of the standardized AI-Workflow IR intermediate representation, constructing a calculation dependency graph, obtaining key path nodes in the calculation dependency graph through a CPA algorithm, forming a key path, marking sub-graphs in the key path, and calculating the workflow of the standardized AI-Workflow IR intermediate representation. Subgraph segmentation and subgraph optimization are carried out on the workflow represented in the standardized AI-Workflow IR by adopting a GRR algorithm, and Agent matching scheduling is carried out on subgraphs based on a deep reinforcement learning algorithm; according to the method, workflow conversion of different frameworks is supported, the scheduling strategy is optimized along with environment changes, the JIT-Agent generates special and precise services according to needs, and reasonable workflow distribution is achieved.

The invention provides a scheduling test method based on a vehicle and road cloud scheduling system and an electronic device, the method is applied to a cloud side device in the vehicle and road cloud scheduling system, and the method comprises the following steps: in response to a scheduling test instruction, receiving to-be-tested vehicle information, to-be-tested object information and a test type carried in the scheduling test instruction; according to the to-be-tested vehicle information and the to-be-tested object information, calling a scheduling model to generate to-be-tested scheduling information; performing a corresponding scheduling test on the to-be-tested scheduling information according to the test type to obtain a test result corresponding to the test type; when it is determined that the to-be-tested scheduling information is abnormal according to the test result, updating the scheduling model to obtain an updated scheduling model; and according to the to-be-tested vehicle information and the to-be-tested object information, calling the updated scheduling model to generate new to-be-tested scheduling information, and carrying out scheduling test on the new to-be-tested scheduling information according to the test type until a test result of the new to-be-tested scheduling information does not have abnormity.

The specification discloses a multi-screen content scheduling system based on multi-dimensional behavior data, which comprises an edge computing device and a cloud processing module. The edge computing device is configured on the side of the screen and comprises a data acquisition and preprocessing unit and a scheduling execution unit. The data acquisition and preprocessing unit acquires multi-dimensional behavior data on the spot of each screen, performs preprocessing and constructs a behavior feature vector. The scheduling execution unit receives and executes a content resource scheduling strategy issued by the cloud processing module. The cloud processing module comprises a cloud database unit and a cloud scheduling unit. The cloud database unit stores data. The cloud scheduling unit performs calculation based on a rule engine and a reinforcement learning network model according to the behavior feature vector, generates a content resource scheduling strategy containing a play content list of each screen, adopts a terminal and cloud cooperation architecture, realizes unified resource scheduling of multiple screens according to different behavior data, and realizes content resource scheduling across time periods and devices.

A method for scheduling edge nodes and a cloud scheduling platform are disclosed, relating to the field of computer technology. This method is applied to a cloud scheduling platform. After receiving a DNS request from a target user, the cloud scheduling platform determines the IP address of the first edge node corresponding to the target access information indicated in the DNS request, based on edge node scheduling information, and returns the IP address of the first edge node to the target user. Since the first edge node scheduled by the cloud scheduling platform has fewer ASs on its network link to the target AS indicated by the target access information than other edge nodes, when the target user accesses the first edge node corresponding to the target domain name through the target AS, the number of AS hops on the network link is reduced compared to accessing other edge nodes corresponding to the target domain name. This reduces access latency and improves network communication quality.

The embodiment of the invention provides a resource scheduling method and system based on a deep learning model. The method comprises the following steps: firstly, relying on a pre-deployed edge intelligent node, acquiring original data through a data collector configured by the edge intelligent node, and executing localization preprocessing on the original data by utilizing the computing power of the edge intelligent node to generate target data, so as to reduce the interference of data transmission redundancy and invalid information; then the collaborative area module performs model training on the target data by using a reinforcement learning scheduling resource algorithm, and realizes effective integration of multi-source data in combination with a multi-modal data fusion technology to generate fused data; performing data purification and feature extraction on the fused data through a deep learning model to generate a scientific and accurate resource allocation strategy; and finally, the resource scheduling task is executed in combination with the target data, the fusion data, the resource allocation strategy and the load balancing strategy, so that the problems of slow response and low efficiency of traditional central cloud scheduling are effectively solved, and the resource allocation efficiency can be remarkably improved.

The invention discloses a flexible regulation and control system and method based on distributed photovoltaic power generation, and belongs to the technical field of photovoltaic energy storage regulation and control, and the system specifically comprises a power grid cloud scheduling platform, a meteorological sensing module, a load prediction module, a photovoltaic power generation prediction module, a net load evaluation module and an energy storage distribution scheduling module. The method comprises the steps of obtaining a net load curve based on a future generating capacity curve and a future load curve, introducing a net load reference line to compare and mark a charging window and a discharging window, and decomposing the net load curve into a high-frequency component and a low-frequency component by adopting a frequency domain decomposition technology; then different battery scheduling sub-modules are introduced to respond to the high-frequency component and the low-frequency component respectively, an optimal charging and discharging instruction in a future period is formulated based on net load absolute values and battery characteristic parameters under a charging window and a discharging window, and through distributed dynamic flexible energy storage scheduling, rapid charging and discharging response is ensured, and the charging and discharging efficiency is improved. And a peak clipping and valley filling plan can be stably executed.

This invention belongs to the field of digital cloud scheduling technology for road networks, specifically relating to a method and system for digital cloud scheduling of road networks based on vehicle-road cooperation. The method includes: collecting traffic flow data using roadside units and identifying vehicle information; analyzing the vehicle information to obtain the vehicle spatial interference index and vehicle behavior interference index of unconnected vehicles, and combining these to generate a penetration impact index for unconnected vehicles; using the penetration impact index of unconnected vehicles to obtain a green wave attenuation index through green wave traffic interference analysis; using the green wave attenuation index to obtain road saturation through road capacity assessment; and invoking the corresponding green wave coordination strategy based on the road saturation. This invention achieves quantitative and dynamic differentiated scheduling of interference from unconnected vehicles, effectively improving the efficiency of green wave traffic and the overall operational efficiency of the road network in mixed traffic flow environments.

This invention discloses a method and system for vehicle access edge cloud scheduling based on multi-dimensional scenarios. The method includes: a scheduling scenario configuration step, which predefines multiple scheduling scenarios and configures corresponding scheduling strategies for each scenario; a vehicle request processing step, which receives and verifies the access request from the vehicle terminal and extracts the vehicle location and service requirements; a scheduling decision step, which determines the target scheduling scenario based on the service requirements and selects the optimal target edge cloud node based on the scenario's strategy, vehicle location, and the real-time resource status of each edge cloud node; and a service redirection step, which returns the target node information to the vehicle, enabling it to directly access and obtain services. This invention achieves intelligent and dynamic vehicle scheduling through scenario-based strategies and multi-factor decision-making, effectively improving the system's ability to cope with tidal loads, ensuring low-latency services, and optimizing resource utilization efficiency.

The invention relates to a method and system for full-scene multi-mode unmanned vehicle patrol on a road, and the method comprises the steps: R1, collecting the high-definition map data of a patrol line and a patrol region through a high-definition map collection vehicle based on the demands of multi-region multi-line unmanned vehicle patrol; and R2, establishing a patrol scheduling map based on the high-definition map data of the patrol area and the cloud scheduling platform, and performing road grading marking on the map area and the path line according to a division standard to form a patrol scheduling dynamic map of multi-level road marking. According to the method, the security and protection monitoring multi-mode patrol requirement is met, free switching of multi-mode patrol in the open road automatic driving unmanned patrol state is achieved, and open road multi-patrol line automatic driving path switching is achieved through regional division path re-planning.

The invention, which relates to the technical field of the Internet of Things, discloses an intelligent management and control system for an Internet of Things device based on cloud collaboration, comprising a cloud control center, an edge gateway and an Internet of Things device terminal which are physically connected. The cloud control center is provided with a dynamic adaptation engine, and the dynamic adaptation engine comprises a multi-source heterogeneous sensing module which collects network transmission indexes and device resource states. According to the Internet of Things equipment intelligent management and control system based on cloud collaboration, by constructing a cloud-edge-equipment three-level collaboration dynamic adaptation engine, the problem that the equipment resource state and the cloud scheduling instruction are disjointed in the network fluctuation environment is effectively solved. The multi-source sensing fusion and neural network decision mechanism is utilized, real-time accurate adjustment of bandwidth allocation, task migration and data compression strategies is achieved, the reliability of instruction delivery and the rationality of equipment resource use under the weak network condition are ensured, and the cooperation efficiency and the overall stability of the system in the complex network environment are improved.

The application relates to the technical field of intelligent traffic control systems, and particularly discloses a cloud scheduling system and method for multi-vehicle cooperative automatic driving, which comprises a cooperative task scheduling module, a global conflict prediction and elimination module, a congestion alleviation decision module and a hierarchical cooperation module. The cooperative task scheduling module allocates transportation tasks based on a multi-objective optimization model; the global conflict prediction and elimination module identifies conflicts in advance and generates minimum disturbance adjustment instructions through a space-time occupation grid tensor; the congestion alleviation decision module identifies congestion and issues formation control instructions by using a traffic flow dynamic model; and the hierarchical cooperation module realizes local negotiation between vehicles and Gaussian path offset avoidance based on a weight potential field function when cloud control is interrupted. The application realizes efficient, safe and robust cooperative operation of multiple automatic driving vehicles in a shield tunnel.