Multi-mode cooperation method and system for multi-network structured industrial system

Abstract

The invention discloses a multi-mode cooperation method and system for a multi-network structured industrial system, and the method comprises the steps: carrying out the modeling according to the multiple relationships between an industrial agent and an agent cluster in different network layers, and enabling the multi-network structured industrial system to be modeled as a multi-layer undirected network; defining an agent and agent cluster load state model and a task model, inputting a task surge amount and updating the load state of the agent bearing the task; in order to flexibly adjust the resource load between network layers, the method dynamically determines the inter-layer cooperative load coefficient by analyzing the heterogeneity of tasks and the real-time load condition of the network layers; and finally, comprehensively considering the collaboration cost and the risk of the key agent, selecting an optimal collaboration strategy, realizing multi-mode collaboration through dynamic interaction among different agents, clusters and network layers, optimizing load distribution and resource scheduling of the system, solving abnormity caused by task surge in the industrial system, and realizing multi-mode collaboration.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-network structured industrial software systems, and mainly relates to a multi-mode collaboration method and system for multi-network structured industrial systems. Background Art

[0002] In recent years, with the rapid development of artificial intelligence, the Internet of Things, and automation technology, the structure of industrial systems has become increasingly complex, showing significant multi-network characteristics. In such a multi-networked industrial system, industrial agents are interconnected through a variety of physical networks and information networks to collaboratively complete data transmission, task coordination, and control operations. This complex and diverse network structure not only improves the intelligence level and overall efficiency of the system, but also brings new challenges to system stability. In a multi-networked industrial system, when the demand for a task suddenly surges, causing the industrial agent or agent cluster that undertakes the task to be overloaded, the abnormal situation of the task and the overload state of the agent will spread in the industrial system network, thereby affecting the operating efficiency of the entire industrial system, and in severe cases may even cause the entire system to fail or collapse.

[0003] In industrial systems, in order to reduce the losses caused by task surges and restore the normal operation of the system as soon as possible, it is necessary to achieve multi-mode collaboration through dynamic interactions between different agents, clusters and network layers. Among the collaborative scheduling methods for industrial systems, most methods focus on establishing collaborative relationships between homogeneous targets, that is, finding agents with exactly the same capabilities in the same network layer. There are also a few methods that determine a fixed inter-layer collaborative load coefficient based on multiple network structures, and all tasks use a fixed coefficient for inter-layer collaboration. However, the above methods ignore the heterogeneity of tasks and network layers, and do not fully utilize the multiple collaborative modes brought by multiple network structures to agents and agent clusters. On the other hand, key agents in industrial system networks play an important role. Failures such as overload of key agents usually affect multiple agent clusters and network layer structures, which can cause a large cascading impact. Therefore, protecting key agents and reducing the risk of key agent failure are of great significance to the stability of industrial system networks. However, existing collaborative scheduling strategies have ignored this important issue. Current methods often only focus on objectives such as cost and load balancing, ignoring the importance of key agents in the network and the stability of the network structure.

[0004] At present, the multi-mode coordination methods for coping with the surge of tasks in multi-network structured industrial systems have the following main problems: (1) Ignoring the multi-mode coordination between agents, clusters and network layers. Most existing methods focus on the coordination between agents with exactly the same capabilities and search for target agents in the same network layer, ignoring the multiple relationships between different network layers and agent clusters in the industrial system, resulting in the industrial network coordination mode and low task completion rate. (2) The coordination load coefficient between network layers is fixed. The existing coordination scheduling methods considering multi-network structures will fix the inter-layer coefficient. When different tasks are coordinated between layers, the fixed coefficient is used. The differences in resource types and demand for different tasks are not taken into account, thus ignoring the heterogeneity of tasks. (3) Lack of protection for key agents and guarantee of network stability. Most existing coordination scheduling methods determine the coordination goals based on factors such as cost and load balancing. They do not consider whether the coordination strategy will bring risks to the target object. They also ignore the important role of key agents in the multi-network structure for the overall network, which makes it difficult to ensure the stability of the network structure. Summary of the invention

[0005] The present invention aims at the problems existing in the prior art and proposes a multi-mode collaboration method and system for multi-network structured industrial systems. Firstly, the multi-network structured industrial system is modeled as a multi-layer undirected network according to the multiple relationships between industrial agents and agent clusters in different network layers. The load state model and task model of agents and agent clusters are defined, the task surge is input and the load state of the agent that undertakes the task is updated. In order to flexibly adjust the resource load between network layers, the method dynamically determines the inter-layer collaboration load coefficient by analyzing the heterogeneity of tasks and the real-time load of the network layer. Finally, the optimal collaboration strategy is selected by comprehensively considering the collaboration cost and the risk of key agents. Multi-mode collaboration is achieved through dynamic interaction between different agents, clusters and network layers, the load distribution and resource scheduling of the system are optimized, the anomalies caused by the surge of tasks in the industrial system are solved, and multi-mode collaboration is achieved.

[0006] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a multi-mode collaboration method for a multi-network structured industrial system, comprising the following steps:

[0007] S1. Modeling multiple relationships between agents and agent clusters: Based on the multiple relationships between industrial agents and agent clusters in different network layers, the multi-network structured industrial system is modeled as a multi-layer undirected network.

[0008] S2. Define the load state model and task model of the agent and agent cluster: the load state model of the agent and agent cluster represents the state of the industrial agent and agent cluster through the current load Load_cur_node, the remaining load Load_ftee_node and the maximum load Load_max_node; the task model models the tasks in multiple networked industrial systems based on the characteristics of the tasks in the real industrial system;

[0009] S3, input task surge: input task surge and update the load state of the agent that undertakes the task;

[0010] S4. Determine the dynamic inter-layer coordination load coefficient: According to the heterogeneity of different tasks and the real-time load of the network layer, evaluate the resource type required for the task, the coordination demand, and the load of the network layer, and dynamically determine the inter-layer coordination load coefficient;

[0011] S5. Determine the collaborative target at the target network layer: select the collaborative target by minimizing the collaborative cost and the risk of the key intelligent agent; the minimized collaborative cost refers to the more edges that the collaborative parties have in the industrial system network; the key intelligent agent risk refers to the ratio of the current load of the key intelligent agent to its maximum load capacity.

[0012] As an improvement of the present invention, in step S1, the multi-network structured industrial system network G = {G1, G2, ..., G l , ..., G L}, the lth layer network is represented as Represents nodes in a multi-network, i.e., agents; Indicates whether there is a connection between the agents in the l-th layer network. The connection matrix between industrial agent i and agent j in the l-th layer network is expressed as

[0013]

[0014] represents the set of agent clusters in the l-th layer network, where K l represents the number of agent clusters in this layer, and each cluster represents the resource set of the task in the l-th layer network, where H l Indicates the number of resource types in this layer.

[0015] As an improvement of the present invention, the load state model of the agent and the agent cluster in step S2 is specifically:

[0016] Load_cur_node l,i，r +Load_free_node l,i,r = Load_max_nodel，i，r

[0017] Load_cur_group l,k，r +Load_free_group l,k,r = Load_max_qroup l,k,r

[0018] Among them, Load_cur_node l,i，r is the current load of resource r of agent i at layer l, Load_free_node l,i，r is the remaining load of resource r of agent i at layer l, Load_max_node l,i，r is the maximum load of resource r of agent i at layer l, Load_cur_qroup l，k，r is the current load of resource r of agent cluster k at layer l, Load_free_group l,k，r is the remaining load of resource r of agent cluster k at layer l, Load_max_group l，k，r is the maximum load of resource r of agent cluster k at layer l.

[0019] As another improvement of the present invention, the load state of the agent cluster undertaking the task in step S3 is specifically: The surge in the number of tasks is added to the task cluster C lk of the current load and reduce C accordingly lk The remaining load:

[0020] Load_cur_group l,k,r ′ = Load_cur_group l,k，r +Q lkr

[0021] Load_free_group l，i,r ′ = Load_free_qroup l,i,r -Q lkr

[0022] Among them, Load_cur_group l,k，r is the load of resource r of agent cluster k at level l before task surge, Load_cur_group l,k,r ′ is the current load of resource r of agent cluster k at the lth layer after task surge.

[0023] As another improvement of the present invention, in step S3, the current load of the resources contained in the agent cluster k at the lth layer after the task surge is compared with the current load of the resources contained in the agent cluster C. lkCompare the maximum load capacity of the agent cluster to determine whether the agent cluster is overloaded after the task surge:

[0024] When the task surges, the current load of the resources contained in the agent cluster k at the lth layer is less than that of cluster C. lk When the maximum load capacity is reached, it indicates that after the task surges, the intelligent cluster k at the lth layer is able to undertake the task, and the industrial network does not have any abnormalities. lk The amount of tasks accepted is the total amount of new tasks;

[0025] When the task surges, the current load of the resources contained in the agent cluster k at the lth layer is greater than that of cluster C lk When the maximum load capacity of cluster C is reached, an abnormality occurs in the industrial network, and the collaborative demand is calculated; the collaborative demand exceeds the maximum load capacity of cluster C lk The parts with the largest capacity are:

[0026] M lkr =Q lkr -Load_free_group l,k, r

[0027] Among them, M lkr is the collaborative demand for resource r in the agent cluster k at the lth layer.

[0028] As another improvement of the present invention, in step S3, when the task surges, the agent cluster k at the lth layer is capable of undertaking the task, and the industrial network does not have any abnormality, the newly added task amount is allocated according to the ratio of the remaining load of the agent members in the cluster, specifically:

[0029]

[0030]

[0031] Among them, Load_cur_node l，i，r is the current load of resource r of agent i at layer l, and agent i is a member of agent cluster k, Load_free_node l,i，r is the residual load of resource r of agent i at layer l.

[0032] As another improvement of the present invention, the step S4 of determining the dynamic inter-layer collaborative load coefficient specifically includes the following steps:

[0033] S41: Calculate the overall load status and average load status of a multi-network structured industrial system:

[0034] Stare l =Load_cur_net l / Load_max_netl

[0035]

[0036] Among them, State l is the network layer load status of the industrial network at layer l, avg State It is the average load status of the entire industrial network;

[0037] S42: Establish a backup industrial system network set Obj_net in multiple networks, by determining the types of resources required for the task Select from multiple industrial system networks All resource types of network layers required as backup network layers;

[0038]

[0039]

[0040] in, is the resource demand type of enterprise group k at level l, The resource types contained in the network set Obj_net;

[0041] S43: Considering the network load balance after multi-mode collaboration, determine the inter-layer collaboration load coefficient α:

[0042]

[0043] Among them, the target network layer number selected from the backup industrial system network set Obj_net is in, State in To implement the target network layer load state after the collaborative strategy is executed, Stare l It is the load status of the source network layer after coordination.

[0044] As a further improvement of the present invention, in step S5, the coordination cost is specifically:

[0045]

[0046] Among them, Cost lky represents the coordination cost between the agent cluster k at the lth layer and the agent cluster y at the lth layer, (acn) ij All the edge relationships for the industrial system network;

[0047] The key agent risks are specifically:

[0048]

[0049] Among them, Risk key_allIt represents the risk of key intelligent agents in the industrial system network, and key_node is the known set of key intelligent agents in the industrial system network structure.

[0050] In order to achieve the above-mentioned purpose, the present invention also adopts a technical solution: a multi-mode collaborative system for multi-network structured industrial systems, including a computer program, which implements the steps of any of the above-mentioned methods when executed by a processor.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] (1) The present invention models multiple relationships between agents and agent clusters in industrial system networks, taking into account that industrial agents may have different relationships in different network layers and clusters. The multiple relationships between agents further lead to collaboration between agent clusters and network layers in multiple networks. The method of the present invention fully considers the above relationships and reasonably models them, and utilizes the characteristics of multiple networked structures to more efficiently perform multi-mode collaboration of multiple networked industrial systems.

[0053] (2) The present invention realizes the adaptive dynamic change of the collaborative load coefficient between network layers in the industrial system network. The method of the present invention takes into account the heterogeneity of different tasks and the real-time load of the network layer, evaluates the resource type and demand required by the task and the load of the network layer, and dynamically determines the collaborative load coefficient between network layers, thereby finding the most suitable collaborative target and determining the optimal collaborative mode and strategy.

[0054] (3) The present invention reduces the risk of failure of key intelligent agents in industrial system networks and improves network stability. In the process of finding collaborative targets, the method of the present invention fully considers the important impact of key intelligent agents on network stability, quantifies the risk value of key intelligent agents in the network, ensures that the collaborative strategy will not cause overload of key intelligent agents, thereby causing significant risks to the system, and effectively enhances the stability of the entire network. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of the overall steps of the method of the present invention;

[0056] Figure 2 is a flowchart of the steps of determining the dynamic inter-layer collaborative load coefficient in step S4 of the method of the present invention;

[0057] Figure 3 It is a flow chart of the steps of determining collaborative targets within the target layer in step S5 of the method of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0059] Example 1

[0060] Based on the problems existing in the current multi-mode collaboration methods in multi-network structured industrial systems, the core goal of the multi-mode collaboration method proposed in the present invention to cope with the surge in tasks is to establish a multi-network structured industrial system network, propose a multi-mode collaboration mechanism, and realize multi-mode collaboration through dynamic interaction between different subjects, clusters and network layers, dynamically adjust the collaboration load coefficient, protect key intelligent agents in the system, optimize the system's load distribution and resource scheduling, thereby solving the abnormal situation of task surge and ensuring the stability and efficiency of the industrial system network.

[0061] There is a total of L layers of multi-networked industrial system network G = {G1, G2, ..., G l , ..., G L}, the lth layer network is represented as Represents the industrial intelligent agent in the industrial system network, Indicates whether there is a connection relationship between the agents in the l-th layer network, represents the set of agent clusters in the l-th layer network, Represents the resource collection of the tasks in the l-th layer network.

[0062] In a multi-networked industrial system, when an industrial agent or agent cluster is performing tasks, there may be a surge in tasks. l There is an emergency in the task undertaken by the agent cluster k, and the task volume surges by Due to the different scales of tasks and the diversity of resources required in industrial systems, It is a collection of various resources. lkr It is represented by the surge in resource r in the enterprise group k at the lth layer. In the face of a surge in tasks in an industrial system, the load of the agent or agent cluster that undertakes the surge task is first updated to calculate the latest current load Load_cur_group l,k,r ′ and the remaining load Load_free_group l,i，r'. Then, check whether the industrial agent or agent cluster is capable of handling this surge in tasks. If the agent or agent cluster that undertakes the surge task is not overloaded, that is, the current load of each resource of the agent plus the load after the surge is still lower than the maximum load that can be carried, then the surge in the task has no impact on the state of the agents in the industrial system, and the industrial network is not abnormal. On the contrary, if the current load of one or more resources of the agent plus the load after the surge exceeds the maximum load that can be carried, an overload occurs. In the face of this kind of industrial system anomaly caused by the surge in tasks, it is necessary to further calculate the types and amounts of resources that need to be coordinated for the task, and provide effective coordination strategies to handle abnormal situations in the industrial system network.

[0063] In order to efficiently solve the abnormal situations of industrial systems caused by the surge in tasks, the method of the present invention will select the optimal collaborative strategy and realize multi-mode collaboration through dynamic interactions between different intelligent agents, clusters and network layers, so as to optimize the load distribution and resource scheduling of the system when dealing with the surge in tasks. First, the collaborative load coefficient between network layers is calculated based on the heterogeneity of tasks and the real-time load of the network layer. Specifically, a backup industrial system network set Obj_net is established in multiple networks, and then the current load and maximum load of each layer are considered in Obj_net. The optimal collaborative network layer is selected with the overall load balancing of the network as the goal, and the inter-layer collaborative load α is dynamically determined. Then, after determining the target network layer and the collaborative load coefficient, the inter-layer demand within the computing layer is further calculated. The resource demand for each layer is (1-α)*Q lkr , the resource demand between layers is α*Q lkr Finally, the agent cluster Obj in the backup layer is established in the target layer. target , comprehensively consider the collaborative cost Cost target Risk of key intelligent agents after collaboration target , select the most efficient industrial intelligent agent or industrial intelligent agent cluster to collaboratively complete tasks and solve the abnormal conditions of the industrial network caused by the surge in tasks.

[0064] like Figure 1 As shown in the figure, the multi-mode collaborative method for coping with task surge in a multi-network structured industrial system specifically includes the following steps:

[0065] Step S1: Considering the different collaborative relationships that may exist among industrial agents in different network layers and clusters, as well as the collaboration between agent clusters and network layers in multiple networks, graph theory techniques are used to model the multi-network structured industrial system as a multi-layer undirected network.

[0066] In the initial stage, a multi-network structured industrial system network model is defined. The multi-network structured industrial system network is represented by G = {G1, G2, ..., Gl , ..., G L}. The l-th layer network is represented as Represents nodes in multiple networks, that is, intelligent agents with independent perception and decision-making capabilities in industrial system networks; Indicates whether there is a connection relationship between the agents in the l-th layer network. The connection matrix between industrial agent i and agent j in the l-th layer network is expressed as:

[0067]

[0068] represents the set of intelligent agent clusters in the lth layer network. Members of the intelligent agent cluster share resources and collaborate to complete tasks in the industrial system. l represents the number of agent clusters in this layer, and each cluster represents the resource set in the layer l network, where H l Indicates the number of resource types in this layer.

[0069] Step S2: Define the industrial agents and agent cluster load state model in the multi-network structured network, and establish the task model in the industrial system.

[0070] The method of the present invention quantifies the status of industrial agents and agent clusters through current load, remaining load and maximum load, which can be expressed as:

[0071] Load_cur_node l,i，r +Load_free_node l,i,r = Load_max_node l,i,r

[0072] Load_cur_group l,k，r +Load_free_group l,k,r =Load_max_group l,k，r

[0073] Among them, Load_cur_node l，i，r is the current load of resource r of agent i at layer l, Load_free_node l，i，r is the remaining load of resource r of agent i at layer l, Load_max_node l,i，r is the maximum load of resource r of agent i at layer l. l,k，r is the current load of resource r of agent cluster k at layer l, Load_free_group l,k，ris the remaining load of resource r of agent cluster k at layer l, Load_max_group l,k，r is the maximum load of resource r of agent cluster k at layer l.

[0074] Based on the characteristics of tasks in real industrial systems, the tasks in multiple networked industrial systems are modeled. The task set in multiple networked industrial systems is expressed as Where M represents the total number of tasks in the industrial system. Each task T n Using triples <pt n , ct n ,qt n > to indicate that pt n Indicates the responsibility of T n The agent or agent cluster of the task, ct n Represents task T n The types of resources required and qt n Represents task T n The amount of resources required.

[0075] Step S3: Input the task surge and update the load status of the agent or agent cluster that undertakes the task.

[0076] Assume that an emergency occurs in the task undertaken by the agent cluster k at the lth layer, and the task scale increases, resulting in a surge in the amount of tasks. It is known that the surge is The surge volume is a collection of multiple resource quantities. lkr It is represented as the surge of resource r in the agent cluster k at the lth layer, where

[0077] Calculate the load status of the agent cluster that undertakes the task. The surge in is added to the cluster C that takes on the task lk of the current load and reduce C accordingly lk The remaining load is obtained. Specifically:

[0078] Load_cur_group l,k,r ′ = Load_cur_group l，k，r +Q lkr

[0079] Load_free_group l，i，r ′=Load_free_group l，i，r -Q lkr

[0080] Among them, Load_cur_group l，k，ris the load of resource r of agent cluster k at level l before task surge, Load_cur_group l，k，r ′ is the current load of resource r of agent cluster k at layer l after task surge, Load_free_group l,i，r Load_free_group is the remaining load of resource r of agent cluster k at layer l before the task surge, that is, the remaining capacity of the resource that the agent cluster can still bear. l，i，r ′ is the remaining load of resource r of agent cluster k at layer l after task surge.

[0081] After the surge in computing tasks, the agent cluster is overloaded. By comparing the current load of the resources contained in the agent cluster k at the lth layer after the surge in tasks with the maximum load capacity of the cluster:

[0082]

[0083] Among them, Load_max_group l，k，r is the maximum load capacity of resource r of agent cluster k at layer l. If the above formula is established, it means that after the task surges, agent cluster k at layer l is still able to undertake the task, and there is no abnormality in the industrial network.

[0084] If the task surge does not cause industrial network abnormalities, then the additional task volume is calculated and the load status of each agent member in the agent cluster is updated. lk If cluster C is able to complete the new tasks and does not affect the industrial network, lk The amount of tasks accepted is the total amount of new tasks:

[0085]

[0086]

[0087] Among them, A lkr is the amount of resource r received by the agent cluster k at the lth layer, M lkr is the collaborative demand for resource r in the agent cluster k at the lth layer. The additional tasks are allocated according to the proportion of the remaining load of the agent members in the cluster:

[0088]

[0089]

[0090] Among them, Load_cur_node l，i，r is the current load of resource r of agent i at layer l, and agent i is a member of agent cluster k, Load_free_nodel,i，r is the residual load of resource r of agent i at layer l.

[0091] If an abnormality occurs in the industrial network, it is necessary to determine the appropriate collaborative strategy to complete the task collaboratively, so the collaborative demand is calculated first. Since the intelligent agent cluster k at the lth layer is overloaded, the calculation exceeds the cluster C lk The largest capacity part:

[0092] M lkr =Q lkr -Load_free_group l,k,r

[0093] Among them, M lkr is the collaborative demand for resource r in the agent cluster k at the lth layer.

[0094] Step S4: Calculate the dynamic collaborative load coefficient, taking into account the heterogeneity of tasks and the real-time load of the network layer, evaluate the resource type and demand required for the task and the load of the network layer, and dynamically determine the collaborative load coefficient between network layers. Figure 2 shown.

[0095] S41: Calculate the overall load status and average load status of the multi-network structured industrial system. The overall load status of the industrial system network is calculated by calculating the overall current load of each layer of the network and the maximum load capacity of the industrial intelligent body in the network layer. It is expressed as:

[0096]

[0097]

[0098] State l =Load_cur_net l / Load_max_net l

[0099]

[0100] Among them, Load_cur_net l Load_max_net is the current load of the network layer of the industrial network at layer l. l is the maximum load of the network layer of the industrial network at layer l, K l is the total number of network layers in multiple industrial networks, State l is the network layer load status of the industrial network at layer l, avg State It is the average load status of the entire industrial network.

[0101] S42: Establish a backup industrial system network set Obj_net in multiple networks. By determining the types of resources required for the task Then select the multiple industrial system network that includes The network layer of all resource types required is used as a backup network layer. It is expressed as:

[0102]

[0103]

[0104] in, is the type of resources required by enterprise group k at level l, The resource types contained in the network collection Obj_net.

[0105] S43: Considering the network load balance after executing the collaborative strategy, select the target network layer in the backup industrial system network set and determine the inter-layer collaborative load coefficient. By calculating the network layer load state after executing the multi-mode collaborative strategy, and taking the minimum difference between the average load state of the target network layer and the industrial network after multi-mode collaboration as the goal, the optimal network layer and the specific value of α are calculated. Specifically:

[0106]

[0107]

[0108]

[0109] Among them, the target network layer number selected from the backup industrial system network set Obj_net is in, State in The target network layer load state after inter-layer coordination, State l It is the load status of the source network layer after coordination.

[0110] Step S5: Find the collaborative target in the target network layer, such as Figure 3 After finding the target network layer, it is necessary to find a specific agent or agent cluster in the target layer and the source network layer as the collaborative target.

[0111] First, calculate the cost of the collaboration strategy. The cost of multi-mode collaboration is defined based on the relationship between the targets in the industrial system. That is, the more edges the two parties in the collaboration have in the industrial system network, the closer the relationship between the targets is, and the lower the collaboration cost between the two parties. Specifically:

[0112]

[0113] Among them, Cost lkyrepresents the coordination cost between the agent cluster k at the lth layer and the agent cluster y at the lth layer, (acn) ij All edge relationships of the industrial system network.

[0114] Secondly, the risk of key intelligent entities in the industrial system network is calculated. The risk of key intelligent entities is quantified by calculating the ratio of the current load of the key intelligent entity to its maximum load capacity. The larger the calculated ratio, the greater the risk of failure of the key intelligent entity. Specifically:

[0115]

[0116] Among them, Risk key_all It represents the risk of key intelligent agents in the industrial system network, and key_node is the key intelligent agent in the input industrial system network structure.

[0117] Finally, find the optimal collaborative target. Select the collaborative target by minimizing the collaborative cost and the risk of the key agent. Specifically:

[0118] min(a*Cost target +b*Risk target ).

[0119] In summary, the present invention discloses a multi-mode collaboration method for multi-network structured industrial systems, especially in response to a surge in tasks, by achieving multi-mode collaboration through dynamic interactions between different agents, clusters, and network layers, thereby optimizing the system's load distribution and resource scheduling when responding to a surge in tasks, and solving the anomalies caused by a surge in tasks in the industrial system. Different from the existing collaborative scheduling methods, the method proposed in the present invention models the relationship between agents and agent clusters in the industrial system network, fully considering the multiple relationships of industrial agents in different network layers and clusters, and the synergy between agent clusters and network layers in multiple networks; considering the heterogeneity of different tasks and the real-time load of the network layer, the adaptive change of the collaborative load coefficient between network layers is achieved.

[0120] In addition, the present invention quantifies the risks of key intelligent entities in the industrial system network and effectively enhances the stability of the industrial system network structure.

[0121] It should be noted that the above content only illustrates the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications all fall within the protection scope of the claims of the present invention.

Claims

1. A multi-mode collaborative method for multi-network structured industrial systems, characterized in that: The steps include: S1. Modeling multiple relationships between agents and agent clusters: Based on the multiple relationships between industrial agents and agent clusters in different network layers, the multi-network structured industrial system is modeled as a multi-layer undirected network. S2. Define the load state model and task model of the agent and agent cluster: the load state model of the agent and agent cluster represents the state of the industrial agent and agent cluster through the current load Load_cur_node, the remaining load Load_free_node and the maximum load Load_max_node; the task model models the tasks in multiple networked industrial systems based on the characteristics of the tasks in the real industrial system; S3, input task surge: input task surge and update the load state of the agent that undertakes the task; S4. Determine the dynamic inter-layer coordination load coefficient: According to the heterogeneity of different tasks and the real-time load of the network layer, evaluate the resource type required for the task, the coordination demand, and the load of the network layer, and dynamically determine the inter-layer coordination load coefficient; S5. Determine the collaborative target at the target network layer: select the collaborative target targt by minimizing the collaborative cost and the risk of the key intelligent agent; the minimized collaborative cost refers to the more edges that the collaborative parties have in the industrial system network; the key intelligent agent risk refers to the ratio of the current load of the key intelligent agent to its maximum load capacity.

2. The multi-mode collaboration method for a multi-network structured industrial system according to claim 1, characterized in that: In step S1, the multi-network structured industrial system network G={G1, G2, ..., G l ,…,G L }, the lth layer network is represented as Represents nodes in a multi-network, i.e., agents; Indicates whether there is a connection between the agents in the l-th layer network. The connection matrix between industrial agent i and agent j in the l-th layer network is expressed as represents the set of agent clusters in the l-th layer network, where K l represents the number of agent clusters in this layer, and each cluster represents the resource set of the task in the l-th layer network, where H l Indicates the number of resource types in this layer.

3. The multi-mode collaboration method for a multi-network structured industrial system according to claim 2, characterized in that: The load state model of the agent and agent cluster in step S2 is specifically: Load_cur_node l,i,r +Load_free_node l,i,r =Load_max_node l,i,r Load_cur_group l,k,r +Load_free_group l,k,r =Load_max_group l,k,r Among them, Load_cur_node l,i,r is the current load of resource r of agent i at layer l, Load_free_node l,i,r is the remaining load of resource r of agent i at layer l, Load_max_node l,i,r is the maximum load of resource r of agent i at layer l, Load_cur_group l,k,r is the current load of resource r of agent cluster k at layer l, Load_free_group l,k,r is the remaining load of resource r of agent cluster k at layer l, Load_max_group l,k,r is the maximum load of resource r of agent cluster k at layer l.

4. The multi-mode collaboration method for a multi-network structured industrial system according to claim 3, characterized in that: The load state of the agent cluster that undertakes the task in step S3 is specifically: The surge in the number of tasks is added to the task cluster C lk of the current load and reduce C accordingly lk The remaining load: Load_cur_group l,k,r ′=Load_cur_group l,k,r +Q lkr Load_free_group l,i,r ′=Load_free_group l,i,r -Q lkr Among them, Load_cur_group l,k,r is the load of resource r of agent cluster k at level l before task surge, Load_cur_group l,k,r ′ is the current load of resource r of agent cluster k at the lth layer after task surge.

5. The multi-mode collaboration method for a multi-network structured industrial system according to claim 4, characterized in that: In step S3, the current load of the resources contained in the agent cluster k at the lth layer after the task surge is compared with the current load of the resources contained in the cluster C. lk Compare the maximum load capacity of the agent cluster to determine whether the agent cluster is overloaded after the task surge: When the task surges, the current load of the resources contained in the agent cluster k at the lth layer is less than that of cluster C. lk When the maximum load capacity is reached, it indicates that after the task surges, the intelligent cluster k at the lth layer is able to undertake the task, and the industrial network does not have any abnormalities. lk The amount of tasks accepted is the total amount of new tasks; When the task surges, the current load of the resources contained in the agent cluster k at the lth layer is greater than that of cluster C lk When the maximum load capacity is reached, the industrial network becomes abnormal, the coordination demand is calculated, and the coordination strategy is determined; The collaborative demand exceeds cluster C lk The parts with the largest capacity are: M lkr =Q lkr -Load_free_group l,k,r Among them, M lkr is the collaborative demand for resource r in the agent cluster k at the lth layer.

6. The multi-mode collaboration method for a multi-network structured industrial system according to claim 5, characterized in that: In step S3, when the task surges, the agent cluster k at the lth layer is capable of taking on the task, and the industrial network does not have any abnormality, the newly added task amount is allocated according to the ratio of the remaining load of the agent members in the cluster, specifically: Among them, Load_cur_node l,i,r is the current load of resource r of agent i at layer l, and agent i is a member of agent cluster k, Load_free_npde l,i,r is the residual load of resource r of agent i at layer l.

7. The multi-mode collaboration method for a multi-network structured industrial system according to claim 6, characterized in that: The step S4 of determining the dynamic inter-layer collaborative load coefficient specifically includes the following steps: S41: Calculate the overall load status and average load status of a multi-network structured industrial system: State l =Load_cur_net l / Load_max_net l Among them, State l is the network layer load status of the industrial network at layer l, avg State It is the average load status of the entire industrial network; S42: Establish a backup industrial system network set Obj_net in multiple networks, by determining the types of resources required for the task Select from multiple industrial system networks All resource types of network layers required as backup network layers; in, is the resource demand type of enterprise group k at level l, The resource types contained in the network set Obj_net; S43: Considering the network load balance after multi-mode coordination, determine the inter-layer coordination load coefficient α: Among them, the target network layer number selected from the backup industrial system network set Obj_net is in, State in State is the target network layer load state after executing the collaborative strategy. l It is the load status of the source network layer after coordination.

8. The multi-mode collaboration method for a multi-network structured industrial system according to claim 7, characterized in that: In step S5, the coordination cost is specifically: Among them, Cost lky represents the coordination cost between the agent cluster k at the lth layer and the agent cluster y at the lth layer, (acn) ij All the edge relationships for the industrial system network; The key agent risks are specifically: Among them, Risk key_all It represents the risk of key intelligent agents in the industrial system network, and key_node is the known set of key intelligent agents in the industrial system network structure.

9. A multi-mode collaborative system for a multi-network structured industrial system, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of any of the above methods are implemented.

Images