Method for generating regional equipment support scheduling scheme based on structured data

Abstract

The invention discloses a method for generating a regional equipment guarantee scheduling scheme based on structured data, and belongs to the technical field of equipment guarantee, and the method comprises the steps: building structural models of an equipment system and a guarantee system through taking task-oriented regional equipment guarantee as a background, and then obtaining supply and demand resource data sets of two structures; then establishing a region guarantee scheduling coarse-grained model and a region guarantee scheduling fine-grained model, and respectively obtaining a coarse-grained model guarantee scheduling scheme T and a fine-grained model guarantee scheduling scheme Tij according to the obtained region guarantee scheduling coarse-grained model and the region guarantee scheduling fine-grained model; and combining the coarse-grained model guarantee scheduling scheme T and the fine-grained model guarantee scheduling scheme Tij to finally obtain a regional equipment guarantee scheduling scheme. According to the regional guarantee scheduling scheme provided by the invention, the scheduling scheme is guaranteed to have robustness under the conditions of a tree structure and a complex scale.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment support, and in particular relates to a method for generating a regional equipment support scheduling plan based on structured data. Background Art

[0002] Equipment support is an activity that uses technology and management means to enable equipment systems to maintain and restore specified functions. In the operational implementation phase, equipment support capabilities are mainly reflected in the ability to dispatch support resources. Under joint combat conditions, multi-service joint and cross-regional operations have become the norm, and the "region-oriented" characteristics of equipment support are significant. In regional operations, there are two types of entities: combat equipment and support units. Each of them constitutes a regional equipment system and a regional support system in a modular hierarchical structure. The equipment system proposes resource requirements, and the support system supplies the required resources. Regional support is to achieve precise matching of resources between the two systems. Therefore, the precise matching of system resources directly affects the support capability and even determines the outcome of regional combat.

[0003] Resource matching is based on the distribution of supply and demand resources, and establishes appropriate matching of supply and demand resources in time and space. According to the complexity of the system, resource matching can be divided into three types: point matching, system matching, and system matching. The research on equipment support resource point matching focuses on how supply points and demand points can achieve the expected spare parts utilization rate and spare parts satisfaction rate under the constraints of resource quantity and geographical location. Wang Wei et al. established a multi-layer planning optimization model for the emergency transportation of ordnance at a single demand point. Wang Suxin et al. considered the dynamic change constraints of vehicle loading capacity, the constraints that nodes can be visited multiple times, and the constraints that demand can be split and transported for the problem of simultaneous delivery of goods to multiple demand points, and established a vehicle path model that considers demand splitting and transportation. Qi Xiaogang et al. established a dynamic maintenance resource optimization scheduling model for multiple supply centers and multiple demand points in different combat stages, so that multiple supply centers can timely and efficiently schedule maintenance resources for demand points, reducing resource scheduling time and the amount of unsatisfied maintenance resources for each demand point. In point matching, supply points gradually develop into supply centers. The research on equipment support resource system matching focuses on how the support system can achieve system goals such as mission success rate, combat readiness, and low economic cost under the constraints of resource quantity, geographical location, task relationship, and support relationship. Wang Yazhou and others believe that the troops need to determine the amount of maintenance resources required by the demand point based on the specific maintenance task volume, and how to reasonably and effectively dispatch maintenance resources to each demand point quickly and in a timely manner, and then study the equipment maintenance support resource prediction and scheduling optimization plan. Li Feifei and others believe that the traditional centralized decision-making method for resource-constrained multi-project scheduling can no longer meet the needs of multi-project management in an information asymmetric environment. The distributed resource-constrained multi-project scheduling problem studies the coordination and allocation decisions of global resources and the scheduling optimization decisions of local projects by multiple autonomous decision makers. The system matching strategy has gradually developed from centralized to distributed. Equipment support resource system matching is to find the best matching solution around the resource characteristics, structural characteristics, spatiotemporal characteristics and evolution laws of the nodes in the two systems to achieve the system mission objectives. Xing Biao et al. introduced complex network theory based on SysML design language to construct an equipment support system architecture view model. Yan Xu et al. believed that it is necessary to establish a support system that matches the equipment system to achieve precise support. Gao Long et al. proposed an improved equipment support system optimization evolution model based on weighted network in view of the fact that the existing equipment support system evolution research based on complex networks did not consider the impact of support tasks on the system evolution process and the evolution model did not fully consider the support capability. Wang Cong, Xu Yaoyao et al. proposed specific key node identification and node attribute matching methods based on support system network modeling, and conducted an in-depth analysis of the support system architecture, but lacked quantitative research on specific support systems from the perspective of obtaining support solutions.The above three types of research are respectively applicable to simple unstructured systems, simple structured systems, and complex hierarchical systems. Their common point is that they are matched according to nodes, but their differences are that the node attributes and node relationships are different. In terms of node attributes, from simply considering quantity and geographical location to task relationships and support relationships, and then to exploring the system hierarchy and network structure, the node attributes are becoming more and more abundant. In terms of node relationships, from independent nodes to simple relationship nodes, and then to nodes with complex associations in the system, the node relationships are becoming more and more complex. By increasing the complexity of node attributes and node relationships, the effectiveness of resource matching is improved. However, there are three problems in the above research: (1) There is theoretical exploration in resource system matching, but there is a lack of quantitative research on specific support systems; (2) The node attributes increase in dimensions, but they are still analyzed independently in each dimension, and no node attribute fusion is performed. At the same time, node attributes and node relationships are also analyzed independently, and the two are not considered in a fusion manner; (3) There is still a lack of quantitative comparative analysis of the impact of resource distribution differences, system architecture differences, support relationship differences, and node scale differences on equipment support resource matching.

[0004] From the above analysis, we can see that, based on the idea of ​​resource system matching, regional equipment support is a comprehensive match of resources in a certain region at both supply and demand nodes. It is necessary to consider not only the demand structure and resource characteristics of individual equipment, but also the system architecture information and system node characteristics. According to the requirements of regional tasks, the overall distribution of resource supply and demand is grasped, and then the configuration and scheduling of regional support resources are coordinated. From a phenomenological point of view, support is the flow of resources between supply and demand. In essence, it is the data transmission or data connection of resources. Therefore, the essence of the regional equipment support problem is the optimal transmission of structured data.

[0005] The optimal transmission of structured data is to connect the two structured data sets at a certain level. Xu Zongben and others believe that the fundamental reason why the two data sets need to be "connected" or can be "connected" is that there are certain "common features" or "invariants" between them. "Maintaining invariants" is the most essential constraint for structured data transmission. In regional equipment support, the resources required by the equipment system and the resources supplied by the support system remain unchanged during the data transmission process, and some structural characteristics of the corresponding resources remain unchanged. The structural characteristics of resources include two aspects: node characteristics and node relationships. Node characteristics are the resource attributes of nodes, and node relationships are the resource relationships of nodes. Feature invariance refers to the fact that the types and quantities of supply and demand resources remain unchanged during the transmission from one node to another. For example, a certain equipment node in the demand system needs spare part A, and a certain support node in the supply system provides spare part A, and the types and quantities of resources remain unchanged; structural invariance refers to the fact that the types and quantities of supply and demand resources remain unchanged during the transmission from one node to another. For example, some resources in the demand system correspond to a certain equipment and some resources in the supply system correspond to a certain support unit, and the structure and relationship of resources remain unchanged. For a long time, people have been studying learning from structured data, such as objects that combine feature and structural information. This kind of graph data is usually a collection of nodes with attributes (multidimensional space vectors) connected by some specific relationship. Structured data objects include time series, trees or images. The optimal transfer theory has inspired many breakthroughs in the field of machine learning, but its basic distance formula cannot utilize the structural information of the object and only relies on the cost function of comparing its feature representation. Peyr′e et al. proposed a comparison method for comparing two distance matrices. These two matrices can be regarded as representations of certain object structures, but only encode the intrinsic structural information. The optimal transfer (OT) distance does not solve the problem of integrating feature and structural information. In recent years, some attempts have been made to define an OT distance by taking the structure as a regularization term and combining the Lagrangian formula of the signal and its temporal structure information. These methods assume that the feature and structural information are in the same space, which is not universal. Nikolentzos et al. proposed a graph similarity measure for discrete labeled graphs with OT. Titouan Vayer et al. further proposed a universal framework that considers feature and structural information and uses a trade-off parameter to balance the importance of features and structure. In fact, regional equipment support has a lot of characteristics and structural information. At the same time, it is difficult to obtain information in regional operations. We should combine the characteristics of regional operations, carry out integration, and design corresponding parameters according to the characteristics of information to flexibly adjust the model. However, the above studies have not achieved a good integration of the characteristics and structural information of system nodes, nor can they adjust the model well according to the parameters. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for generating a regional equipment support scheduling plan based on structured data.

[0007] To achieve the above object, the technical solution adopted by the present invention is:

[0008] A method for generating a regional equipment support scheduling plan based on structured data comprises the following steps:

[0009] S1. Establish regional equipment system structure and regional support system structure;

[0010] S2. Acquire a first supply and demand resource data set and a second supply and demand resource data set according to the regional equipment system structure and the regional support system structure established in step S1;

[0011] S3. A regional security scheduling coarse-grained model is established based on the first supply and demand resource data set obtained in step S2. The regional security scheduling coarse-grained model is:

[0012]

[0013] Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i,j is the supply and demand resource matching between node i and node j, T k，l is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0014] S4. Based on the second supply and demand resource data set obtained in step S2 and the regional security scheduling coarse-grained model obtained in step S3, a regional security scheduling fine-grained model is established. The regional security scheduling fine-grained model is:

[0015]

[0016] in, is the objective function of the fine-grained model, α represents the trade-off parameter between node features and node structure, q represents the distance parameter, and a i and b j They are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, is the supply and demand resource matching between node i and node j, is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, The distance between node i and node k in the equipment architecture and support architecture is the result of integrating the distance within the structure. h is the result of integrating the distance between node j and node l in the equipment architecture and support architecture into the distance within the structure. ik and g jl They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0017] S5. According to the regional security scheduling coarse-grained model obtained in step S3 and the regional security scheduling fine-grained model obtained in step S4, the coarse-grained model security scheduling scheme T and the fine-grained model security scheduling scheme T are obtained respectively. ij ;

[0018] S6: Combine the coarse-grained model guarantee scheduling solution T and the fine-grained model guarantee scheduling solution T obtained in step S5 ij , and obtain the regional equipment support and dispatch plan.

[0019] Preferably, step S2 comprises the following steps:

[0020] S21, assuming that there are m support agencies and n support objects in the regional equipment architecture and regional support architecture established in step S1, and the support agency set is {X1, X2…X m}, the distribution of resources provided by the security agency is {(x1,u1)…(x i ,u i )…(x m ,u m )},x i and u i X i The location and quantity of resources; the set of guaranteed objects is {Y1, Y2…Y n}, the resource distribution required by the protection object is {(y1,v1)…(yj ,v j )…(y n ,v n )},y j and v j Y j Location and quantity of resources; resources from X i Transport to Y j The number is T ij ;

[0021] S22. Based on step S21, the supply distribution of resources is obtained as follows: The resource demand distribution is is a Dirac function, then the obtained resource supply distribution u and demand distribution v are the first supply and demand resource data set;

[0022] S23. Based on step S21, it is assumed that the security institution X i It is a protection unit A collection of security agencies X i The resource distribution of supply is x ik and h ik X ik The location and quantity of resources, h i For X i Resource supply vector; security object Y j It is an equipment unit A collection of protection objects Y j The required resource distribution is y jl and g jl Y jl The location and quantity of resources, g j Y j The resource demand vector, resources from X ik Transport to y jl The number of

[0023] S24. Based on step S23, the supply distribution of resources is obtained as follows: The resource demand distribution is is a Dirac function, then the resource supply distribution h i and demand distribution g j , which is the second supply and demand resource data set.

[0024] Preferably, step S3 comprises the following steps:

[0025] S31. According to the first supply and demand resource data set obtained in step S2, an optimal transmission model for regional security scheduling is obtained. The optimal transmission model for regional security scheduling is:

[0026]

[0027]

[0028] Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the supply and demand resource matching between node i and node j, T k，l is the matching amount of supply and demand resources between node k and node l, d(a i , b j ) is the node characteristic distance, C1(i, k) is the distance between node i and node k in the equipment system architecture and support system architecture, C2(j, l) is the distance between node j and node l in the equipment system architecture and support system architecture, u i and v j are the importance information of equipment architecture and support architecture, respectively, i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0029] S32: Based on the optimal transmission model for regional security scheduling obtained in step S31, the distance between structures d_uv is integrated into the node characteristic distance d(a i , b j ), the new d′(a i , b j ), the distance d within the structure uu =d(x i , x k ) p and d vv =d(x j , x l ) p Integrate them into C1(i, k) and C2(j, l) respectively, and get the new and Then we get the coarse-grained model of regional security dispatch, which is:

[0030]

[0031] Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and ai and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the supply and demand resource matching between node i and node j, T k，l is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively.

[0032] Preferably, step S5 comprises the following steps:

[0033] S51: construct the fine-grained adjacency matrix of the regional equipment architecture and regional support architecture established in step S1 and

[0034] S52, to the fine-grained adjacency matrix obtained in step S51 and Enter the node data x of the regional equipment architecture and regional support architecture respectively ik 、h ik and jl , g jl , normalize the node data h ik and g jl , and obtain the probability vector ph of each supply and demand point i and pg j ;

[0035] S53, combining the regional security scheduling fine-grained model obtained in step S4 with the probability vector ph obtained in step S52 i and pg j Calculate X i To Y j The distribution distance And the guaranteed scheduling probability scheme PT ij ;

[0036] S54, constructing coarse-grained adjacency matrices C1 and C2 of the regional equipment architecture and regional support architecture established in step S1;

[0037] S55, input the node data x of the regional equipment system structure and the regional support system structure into the coarse-grained adjacency matrices C1 and C2 obtained in step S54 respectively. i 、u i and j 、v j , normalize the node data u i and v j , get the probability vectors pu and pv;

[0038] S56: Calculate X based on the regional security scheduling coarse-grained model obtained in step S3 and the probability vectors pu and pv obtained in step S55. ik To Y jl The distribution distance FGW q，α (u, v) and the guaranteed scheduling probability scheme PT;

[0039] S57: The guaranteed scheduling probability scheme PT obtained in step S53 ij Calculate the fine-grained model to ensure the scheduling solution T ij ;

[0040] S58. Calculate the coarse-grained model guarantee scheduling scheme T according to the guarantee scheduling probability scheme PT obtained in step S56.

[0041] Preferably, step S57 includes the following steps:

[0042] S571: According to the guaranteed scheduling probability scheme PT obtained in step S53 ij , get the minimum resource allocation solution And the most resource delivery solution

[0043] S572: Based on step S571, if sum(h i )≥sum(g j ), at this time, the demand can be met, and the fine-grained model guarantees the scheduling solution T ij Minimum resource allocation solution

[0044] S573. Based on step S571, if sum(h i )<sum(g j ), at this time, the demand is partially met, then the fine-grained model guarantees the scheduling solution T ij Contains the most resource delivery solutions and scheduling scheme Two parts.

[0045] Preferably, step S58 includes the following steps:

[0046] S581, according to the guaranteed scheduling probability scheme PT obtained in step S56, obtain the minimum resource allocation scheme T 1 =PT*sum(v) and the maximum resource delivery plan T 2 =PT*sum(u);

[0047] S582: Based on step S581, if sum(u)≥sum(v), the demand can be met, and the coarse-grained model guarantees that the scheduling solution T is the minimum resource configuration solution T. 1 ;

[0048] S583: Based on step S581, if sum(u)<sum(v), the demand is partially met, and the coarse-grained model guarantees that the scheduling solution T contains the optimal guarantee solution and contains the most resource delivery solutions T. 2 and scheduling scheme sum(T 1 -T 2 , 2).

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] (1) The nodes of the regional equipment architecture and regional support architecture established by the present invention contain two types of information: features and structures. Information fusion is performed at the node level, and matching is performed based on features and structural constraints at the system level, thereby transforming the regional support scheduling problem into how to achieve structured data set matching of two systems in the region in an optimal way;

[0051] (2) The regional security dispatch coarse-grained model and the regional security dispatch fine-grained model established by the present invention have the characteristics of structural consistency and convenient FGW distance conversion, so that the model can be switched freely when the resolution is converted;

[0052] (3) The regional security scheduling scheme provided by the present invention is robust under the tree structure and complex scale. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A modular hierarchical structure of a regional equipment architecture in an embodiment of the present invention;

[0054] Figure 2 A modular hierarchical structure of the regional security architecture in an embodiment of the present invention;

[0055] Figure 3 Represent models for structured data;

[0056] Figure 4 Optimal transmission model for structured data scheduling for regional security;

[0057] Figure 5The structure and importance of regional equipment groups;

[0058] Figure 6 To ensure the structure and importance of the regional security organization;

[0059] Figure 7 Optimize process graphs for coarse-grained models;

[0060] Figure 8 It is the error diagram of the guarantee scheme of resource A under different resource distribution under small-scale conditions;

[0061] Fig. 9 Error diagram of the guarantee scheme for resource A when there are three types of resources;

[0062] Fig.10 It is the error diagram of the guarantee scheme of different system structure resources A under medium-scale conditions;

[0063] Fig.11 This is the error diagram of the guarantee scheme when alpha changes;

[0064] Fig.12 This is a comprehensive analysis diagram of the protection plan error based on supply and demand distribution. DETAILED DESCRIPTION

[0065] The following is a description of the embodiments of the present invention. Figures 1 to 12 , the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] The embodiment of the present invention provides a method for generating a regional equipment support scheduling plan based on structured data, which specifically includes the following steps:

[0067] S1. Establish regional equipment system structure and regional support system structure;

[0068] (1) Regional support modular layered structure

[0069] In regional missions, equipment groups perform combat missions, and support agencies complete support missions. Support missions are functional status restoration work performed due to equipment failures and damages during combat missions. Regional support scheduling is to achieve resource supply and demand matching between the support system and the equipment system under certain constraints. The regional equipment system architecture constructed in the embodiment of the present invention includes several equipment groups. The equipment group includes several combat equipment. The equipment is composed of a system. There are different devices in the system. The failure and damage of the equipment require corresponding support resources. Groups, equipment, systems, and equipment have both hierarchical characteristics in terms of composition structure and modular characteristics in terms of functional division. There is also a certain correlation between elements at the same level and between levels. Therefore, the regional equipment system architecture constructed in the embodiment of the present invention is a typical modular hierarchical structure. The specific structure is as follows: Figure 1 Similarly, the regional security system structure constructed in the embodiment of the present invention includes several security agencies, the security agencies include several security bases, the security bases are composed of teams, the teams have different security units, and the security units contain corresponding security resources. The regional security system structure constructed in the embodiment of the present invention is also a typical modular hierarchical structure. The specific structure is as follows Figure 2 shown.

[0070] At the same time, Figure 1 and Figure 2 In the modular hierarchical structures of the two architectures, the system nodes also have importance and geographical location attributes, and the node resources have not only type and quantity attributes, but also time and probability attributes. Therefore, the data in the two architectures are structured data with characteristic attributes and hierarchical structures.

[0071] (2) Regional support structured data expression

[0072] Structured data expressions can be abstracted as follows Figure 3 The model shown, where the circle color represents the node feature information, the circle size represents the node importance information, and the connection relationship represents the node structure information. Feature information Structural information but Figure 3 The structured data of is comprehensively expressed as a probability measure, namely Among them, a i is the node feature, x i is the node structure, δ is the Dirac function of the node feature or structure, u i For importance information.

[0073] S2. Acquiring a first supply and demand resource data set and a second supply and demand resource data set according to the regional equipment system structure and the regional support system structure established in step S1, including the following steps:

[0074] S21, assuming that there are m support agencies and n support objects in the regional equipment architecture and regional support architecture established in step S1, and the support agency set is {X1, X2…X m}, the distribution of resources provided by the security agency is {(x1,u1)…(x i ,u i )…(x m ,u m )},x i and u i X i The location and quantity of resources; the set of guaranteed objects is {Y1, Y2…Y n}, the resource distribution required by the protection object is {(y1, v1)…(y j , v j )…(y n , v n )},y j and v j Y j Location and quantity of resources; resources from X i Transport to Y j The number is T ij ;

[0075] S22. Based on step S21, from the perspective of regional resource distribution, the resource supply distribution is obtained as follows: The resource demand distribution is is a Dirac function, then the obtained resource supply distribution u and demand distribution v are the first supply and demand resource data set; where, The meanings are as follows.

[0076]

[0077] S23, based on step S21, further obtain the expanded high-resolution information: the location and resource quantity of all support units of the support organization, the location and resource quantity of all equipment units of the support object; assuming that the support organization X i It is a protection unit A collection of security agencies X i The resource distribution of supply is x ik and h ik X ik The location and quantity of resources, h i For X i Resource supply vector; security object Y j It is an equipment unit A collection of protection objects Y j The required resource distribution is y jl and gjl Y jl The location and quantity of resources, g j Y j The resource demand vector of X ik Transport to Y jl The number of

[0078] S24. Based on step S23, the supply distribution of resources is obtained as follows: The resource demand distribution is is the Dirac function, M and N are the total number of support units and equipment units respectively. ik Transport to Y jl The number of Then the resource supply distribution h i and demand distribution g j , which is the second supply and demand resource data set.

[0079] S3. A regional security scheduling coarse-grained model is established based on the first supply and demand resource data set obtained in step S2. The regional security scheduling coarse-grained model is:

[0080]

[0081] Among them, FGW q,α (u,v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i,j is the supply and demand resource matching between node i and node j, T k,l is the supply and demand resource matching between node k and node l, d * (a i ,b j ) is a i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0082] Step S3 specifically includes the following steps:

[0083] S31. According to the first supply and demand resource data set obtained in step S2, an optimal transmission model for regional security scheduling is obtained. The optimal transmission model for regional security scheduling is:

[0084]

[0085] Among them, FGW q,α (u,v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i,j is the supply and demand resource matching between node i and node j, T k,l is the matching amount of supply and demand resources between node k and node l, d(a i ,b j ) is the node characteristic distance, C1(i,k) is the distance between node i and node k in the equipment system architecture and support system architecture, C2(j,l) is the distance between node j and node l in the equipment system architecture and support system architecture, u i and v j are the importance information of equipment architecture and support architecture, respectively, i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0086] according to Figure 3 It can be seen from the structured data representation model that the transmission between two structured data is reflected in the flow of data information, that is, the matching degree. The optimal transmission of structured data is the best matching solution for structured data. According to the above analysis, regional support scheduling is the best match between the two structured data of regional equipment system and regional support system. Based on this, the optimal transmission model of regional support scheduling structured data is established, such as Figure 4 As shown in Figure 1, G1 is the equipment system structure, and G2 is the support system structure.

[0087] The present invention considers two structures and Remember M AB (i,j)=d(a i ,b j ), C1(i,k)=d X (x i ,x k ), C2(j,l)=d Y (y j ,y l ). Here a i and b jare the node feature information of the two structures respectively, x i and j are the node structure information of the two structures respectively, u i and v j is the importance information of the two structures. When only the node feature information is considered, the structured data distance is the distance between the two data distributions, including KL divergence, JS divergence and wasserstein distance. When the node feature information and structure information are considered comprehensively, the structured data distance is the fused gromov-wasserstein (FGW) distance, then:

[0088]

[0089] The above-mentioned optimal transmission model for regional security scheduling is constructed in this way.

[0090] S32, based on the optimal transmission model of regional security scheduling obtained in step S31, since the model considers two structures, tree structure and simple network structure, the structure adopts the adjacency matrix representation method, and the distance between nodes is represented by the minimum number of traversed edges between two nodes. The tree structure is as follows Figure 1 and Figure 2 The structure shown in the figure is a simple network structure that considers that all nodes have a connection relationship and the connection strength is the same. Regarding the guarantee relationship, consider X i and Y j The guarantee relationship between ij ,θ ij The value set is set to {1e-3,1,1e3}, 1e-3 means no relationship, 1 means relationship, 1e3 means Y j In X i With respect to position distance, there are two types: inter-structure distance and intra-structure distance. Therefore, the inter-structure distance d_uv is integrated into the node feature distance d(a i ,b j ), the new d′(a i ,b j ), the distance d within the structure uu =d(x i ,x k ) p and d vv =d(x j ,x l ) p Integrate them into C1(i,k) and C2(j,l) respectively, and get the new and Define the distance d between structures uv =d(x i ,y j )p , d(x i ,y j )=|x i -y j ), p is a positive integer, usually 1 and 2. Regarding the characteristic distance. The characteristic distance considers the resource characteristic vector, the location distance between structures, and the supply and demand guarantee relationship, that is, d * (a i ,b j )=d′(a i ,b j )*θ ij Here, the guarantee task reflects the resource supply and demand characteristics of the node, and the regional guarantee scheduling coarse-grained model is obtained. The regional guarantee scheduling coarse-grained model is:

[0091]

[0092] Among them, FGW q,α (u,v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the supply and demand resource matching between node i and node j, T k，l is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively.

[0093] S4. Based on the second supply and demand resource data set obtained in step S2 and the regional security scheduling coarse-grained model obtained in step S3, a regional security scheduling fine-grained model is established. The regional security scheduling fine-grained model is:

[0094]

[0095] in, is the objective function of the fine-grained model, α represents the trade-off parameter between node features and node structure, q represents the distance parameter, and a iand b j They are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, is the supply and demand resource matching between node i and node j, is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, The distance between node i and node k in the equipment architecture and support architecture is the result of integrating the distance within the structure. h is the result of integrating the distance between node j and node l in the equipment architecture and support architecture into the distance within the structure. ik and g jl They are the node data of the regional equipment architecture and the regional support architecture respectively;

[0096] The fine-grained regional security scheduling model transforms the node data u in the coarse-grained regional security scheduling model into i and v j Replaced with the node data h in the second supply and demand resource dataset ik and g jl And get.

[0097] S5. According to the regional security scheduling coarse-grained model obtained in step S3 and the regional security scheduling fine-grained model obtained in step S4, the coarse-grained model security scheduling scheme T and the fine-grained model security scheduling scheme T are obtained respectively. ij , specifically including the following steps:

[0098] S51: construct the fine-grained adjacency matrix of the regional equipment architecture and regional support architecture established in step S1 and

[0099] S52, to the fine-grained adjacency matrix obtained in step S51 and Enter the node data x of the regional equipment architecture and regional support architecture respectively ik 、h ik and jl , g jl , normalize the node data h ik and g jl , and obtain the probability vector ph of each supply and demand point i and pg j ;

[0100] S53, combining the regional security scheduling fine-grained model obtained in step S4 with the probability vector ph obtained in step S52 i and pg j Calculate X i To Y j The distribution distance And the guaranteed scheduling probability scheme PT ij ; Specifically, the node data h in the regional security scheduling fine-grained model ik and g jl Replaced by probability vector ph i and pg j , calculate X i To Y j The distribution distance The distribution distance calculated by The corresponding guaranteed scheduling probability scheme PT can be obtained ij .

[0101] S54, constructing coarse-grained adjacency matrices C1 and C2 of the regional equipment architecture and regional support architecture established in step S1;

[0102] S55, input the node data x of the regional equipment system structure and the regional support system structure into the coarse-grained adjacency matrices C1 and C2 obtained in step S54 respectively. i 、u i and j 、v j , normalize the node data u i and v j , get the probability vectors pu and pv;

[0103] S56: Calculate X based on the regional security scheduling coarse-grained model obtained in step S3 and the probability vectors pu and pv obtained in step S55. ik To Y jl The distribution distance FGW q，α (u, v) and the guaranteed scheduling probability scheme PT; specifically, the node data u in the regional guaranteed scheduling coarse-grained model i and v j Replace them with probability vectors pu and pv respectively, and calculate X ik To Y jl The distribution distance FGW q,α (u,v), the distribution distance FGW obtained by calculation q,α (u,v) can get the corresponding guaranteed scheduling probability plan PT.

[0104] S57: The guaranteed scheduling probability scheme PT obtained in step S53 ij Calculate the fine-grained model to ensure the scheduling solution Tij , including the following steps:

[0105] S571: According to the guaranteed scheduling probability scheme PT obtained in step S53 ij , get the minimum resource allocation solution And the most resource delivery solution Among them, sum(g j ) is the pair vector g j The sum of i ) is the pair vector h i The sum of

[0106] S572: Based on step S571, if sum(h i )≥sum(g j ), at this time, the demand can be met, and the fine-grained model guarantees the scheduling solution T ij Minimum resource allocation solution

[0107] S573. Based on step S571, if sum(h i )<sum(g j ), at this time, the demand is partially met, then the fine-grained model guarantees the scheduling solution T ij Contains the most resource delivery solutions and scheduling scheme Two parts, of which, For the matrix With the matrix After subtraction, sum each row of the resulting matrix;

[0108] S58, calculating the coarse-grained model guarantee scheduling scheme T according to the guarantee scheduling probability scheme PT obtained in step S56, including the following steps:

[0109] S581, according to the guaranteed scheduling probability scheme PT obtained in step S56, obtain the minimum resource allocation scheme T 1 =PT*sum(v) and the maximum resource delivery plan T 2 =PT*sum(u); sum(v) refers to the sum of vector v, and sum(u) refers to the sum of vector u;

[0110] S582: Based on step S581, if sum(u)≥sum(v), the demand can be met, and the coarse-grained model guarantees that the scheduling solution T is the minimum resource configuration solution T. 1 ;

[0111] S583: Based on step S581, if sum(u)<sum(v), the demand is partially met, and the coarse-grained model guarantees that the scheduling solution T contains the optimal guarantee solution and contains the most resource delivery solutions T. 2 and scheduling scheme sum(T 1 -T 2 ,2),sum(T 1 -T 2 ,2) is the matrix T 1 With the matrix T 2 After subtraction, sum each row of the resulting matrix;

[0112] S6: Combine the coarse-grained model guarantee scheduling solution T and the fine-grained model guarantee scheduling solution T obtained in step S5 ij , and obtain the regional equipment support and dispatch plan.

[0113] The regional equipment support scheduling solution provided by the embodiment of the present invention is verified by actual cases.

[0114] 1. Case Background

[0115] Before a mission, according to the plan, there are 4 equipment groups and 20 equipment units to carry out the combat mission, and 5 support organizations and 12 support units to participate in the support mission. The structure and importance of equipment groups and support organizations are as follows: Figure 5 and Figure 6 The information of the equipment group and support organization, such as unit, location, resource supply and demand, is shown in Table 1 and Table 2 below. The support relationship is shown in Table 3 below. The pre-mission support scheduling plan is formulated. The first two data in square brackets in Table 1 and Table 2 are location coordinates, and the last two are the supply and demand quantities of resource A and resource B respectively.

[0116] Table 1 Related information of guarantee institutions and guarantee units

[0117]

[0118] Table 2 Related information of equipment group and equipment unit

[0119]

[0120] Table 3 Support relationship between support organizations and equipment groups

[0121]

[0122] From the results in Table 3, it can be seen that the support relationship between each support organization and all support units and equipment units within the equipment formation is the same, and is the support relationship value of the organization and the formation. For example, the support relationship of X1 and Y1 is 1e-3.

[0123] 2. Calculation results

[0124] In this case, the total supply and demand balance is taken as p = 2, α = 0.5, q = 0.5. The calculation results are as follows. Due to space limitations, only the calculation results of resource A are given. The coarse-grained model guarantees the scheduling plan T as shown in Table 4.

[0125] Table 4 Coarse-grained model guarantee scheduling scheme T obtained in two cases

[0126]

[0127] From the results in Table 4, it can be seen that Scheme 1 and Scheme 2 are the calculation results of the tree structure and simple network structure under the free competition guarantee relationship, respectively. Among them, according to the case data, the guarantee relationship is defined as a fixed combination guarantee relationship, referred to as Relationship 2, and when all matrix elements are adjusted to 1, the guarantee relationship is defined as a free competition guarantee relationship, referred to as Relationship 1, the tree structure is referred to as Structure 1, and the simple network structure is referred to as Structure 2. The abbreviations are used in the following.

[0128] Based on structure 1-relation 1, we get the fine-grained model guarantee scheduling solution T ij , as shown in Table 5 below.

[0129] Table 5 Fine-grained model guarantee scheduling scheme T ij result

[0130]

[0131] 3. Algorithm performance and related discussions

[0132] (1) Algorithm performance analysis

[0133] The coarse-grained model optimization process in the case is as follows Figure 7 As shown in the figure, from the iteration curve, the optimal solution can be found in a relatively small number of iterations, indicating that the algorithm works well. The subsequent related discussions are all based on the coarse-grained model.

[0134] (2) Comparison of different resource distribution functions

[0135] The data of resources A and B in the case are adjusted to different resource distribution functions, as shown in Table 6 below, and the resource distribution difference analysis under small-scale conditions is performed. The calculation results are as follows: Figure 8 shown.

[0136] Table 6 Supply and demand distribution of resources A and B

[0137]

[0138] pass Figure 8It can be seen that when there is an exponential distribution in the supply and demand distribution, the difference in the protection plan is very small. When the supply and demand distribution is a combination of normal distribution and uniform distribution, the protection plan has certain changes. Overall, the plan error is stable within a certain range. The difference at very few points is large, and the normal distribution shows better stability.

[0139] When the number of resource types is increased to three, we get Fig. 9 The results are shown. Fig. 9 The results show that as the types of resources increase, the impact of resource distribution on the security plan is basically the same. Overall, the plan error shows a downward trend, indicating that the model has better robustness when the types of resources increase.

[0140] (3) Comparison of different system structures

[0141] The number of supply and demand nodes in the case is increased by 3 times, which is defined as medium scale. From the analysis of resource distribution differences, it is found that 20 iterations can meet the analysis requirements. Combined with the computer operation efficiency, 20 iterations are selected. The calculation results are as follows: Fig.10 As shown. Fig.10 The results show that the protection scheme has very good stability, and the protection scheme obtained by structure 1 is more stable than that of structure 2.

[0142] (4) Comparison of different trade-off parameters

[0143] Under small-scale conditions, based on structure 1, considering three trade-off parameters α (alpha), the calculation results are shown in Table 7 below. Considering the changes in the protection scheme when alpha changes, the calculation results are as follows Fig.11 As shown in Figure 2. alpha=0 means that only node features are considered, and alpha=1 means that only node structure is considered. Fig.11 The results show that as the alpha value changes, the changes in the safeguard scheme do not show regular changes with alpha, but have a certain degree of stability, indicating that the impact of the trade-off parameters on the safeguard is relatively stable.

[0144] Table 7 Three security solutions when weighing parameters

[0145]

[0146] (5) Comprehensive analysis

[0147] Based on the principle of "supply distribution fixed - demand distribution changing", the calculation results are analyzed under two conditions: small scale and medium scale. Fig.12 shown.

[0148] pass Fig.12The following conclusions can be drawn: the error of the security plan caused by the supply and demand distribution is smaller under medium-scale conditions than under small-scale conditions. With the increase of scale, the cosine distance error and variance error of the plan are significantly reduced. At the same time, it is also found that when the normal distribution and uniform distribution are used, the security plan errors caused by the two distributions gradually approach each other from small scale to medium scale, with different system structures and security relationships. When the exponential distribution is used, the security plan errors caused by the distribution gradually deviate from each other from small scale to medium scale, with different system structures and security relationships.

[0149] In summary, the regional equipment support scheduling scheme provided by the embodiment of the present invention has a certain degree of stability, the sensitivity of the trade-off parameters to the support scheme is within a certain range, the support scheme obtained by the difference in resource distribution is less sensitive to the medium scale than the small scale, and the model performs better when the resource types increase. The support scheme obtained by the difference in system structure and support relationship has similar effects of normal distribution and uniform distribution, the influence of exponential distribution increases with the increase in scale, the tree structure is more stable at different scales, and the influence of support relationship does not show obvious rules.

[0150] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for generating a regional equipment support scheduling plan based on structured data, characterized in that: The following steps are involved: S1. Establish regional equipment system structure and regional support system structure; S2. Acquire a first supply and demand resource data set and a second supply and demand resource data set according to the regional equipment system structure and the regional support system structure established in step S1; S3. A regional security scheduling coarse-grained model is established based on the first supply and demand resource data set obtained in step S2. The regional security scheduling coarse-grained model is: Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the matching amount of supply and demand resources between nodes i and j, T k,l is the supply and demand resource matching between node k and node l, d * (a i ,b j ) is a i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively; S4. Based on the second supply and demand resource data set obtained in step S2 and the regional security scheduling coarse-grained model obtained in step S3, a regional security scheduling fine-grained model is established. The regional security scheduling fine-grained model is: in, is the objective function of the fine-grained model, α represents the trade-off parameter between node features and node structure, q represents the distance parameter, and a i and b j They are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, is the supply and demand resource matching between node i and node j, is the supply and demand resource matching between node k and node l, d * (a i , b j ) is a i and b j The characteristic distance between the distances between structures is integrated, The distance between node i and node k in the equipment architecture and support architecture is the result of integrating the distance within the structure. h is the result of integrating the distance between node j and node l in the equipment architecture and support architecture into the distance within the structure. ik and g jl They are the node data of the regional equipment architecture and the regional support architecture respectively; S5. According to the regional security scheduling coarse-grained model obtained in step S3 and the regional security scheduling fine-grained model obtained in step S4, the coarse-grained model security scheduling scheme T and the fine-grained model security scheduling scheme T are obtained respectively. ij ; S6: Combine the coarse-grained model guarantee scheduling solution T and the fine-grained model guarantee scheduling solution T obtained in step S5 ij , and obtain the regional equipment support and dispatch plan.

2. The method for generating a regional equipment support scheduling plan based on structured data according to claim 1 is characterized in that: Step S2 includes the following steps: S21, assuming that there are m support agencies and n support objects in the regional equipment architecture and regional support architecture established in step S1, and the support agency set is {X1, X2…X m }, the distribution of resources provided by the security agency is {(x1,u1)…(x i ,u i )…(x m ,u m )},x i and u i X i The location and quantity of resources; the set of guaranteed objects is {Y1, Y2…Y n }, the resource distribution required by the protection object is {(y1, v1)…(y j , v j )…(y n , v n )},y j and v j Y j Location and quantity of resources; resources from X i Transport to Y j The number is T ij ; S22. Based on step S21, the supply distribution of resources is obtained as follows: The resource demand distribution is is a Dirac function, then the obtained resource supply distribution u and demand distribution v are the first supply and demand resource data set; S23. Based on step S21, it is assumed that the security institution X i It is a protection unit A collection of security agencies X i The resource distribution of supply is x ik and h ik X ik The location and quantity of resources, h i For X i Resource supply vector of Y; security object Y j It is an equipment unit A collection of protection objects Y j The required resource distribution is y jl and g jl Y jl The location and quantity of resources, g j Y j The resource demand vector, resources from X ik Transport to Y jl The number of S24. Based on step S23, the supply distribution of resources is obtained as follows: The resource demand distribution is is a Dirac function, then the resource supply distribution h i and demand distribution g j , which is the second supply and demand resource data set.

3. The method for generating a regional equipment support scheduling plan based on structured data according to claim 1, characterized in that: Step S3 includes the following steps: S31. According to the first supply and demand resource data set obtained in step S2, an optimal transmission model for regional security scheduling is obtained. The optimal transmission model for regional security scheduling is: Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the matching amount of supply and demand resources between nodes i and j, T k，l is the matching amount of supply and demand resources between node k and node l, d(a i , b j ) is the node characteristic distance, C1(i, k) is the distance between node i and node k in the equipment system architecture and support system architecture, C2(j, l) is the distance between node j and node l in the equipment system architecture and support system architecture, u i and v j are the importance information of equipment architecture and support architecture, respectively, i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively; S32: Based on the optimal transmission model for regional security scheduling obtained in step S31, the distance between structures d_uv is integrated into the node characteristic distance d(a i , b j ), the new d′(a i , b j ), the distance d within the structure uu =d(x i , x k ) p and d vv =d(x j , x l ) p Integrate them into C1(i, k) and C2(j, l) respectively, and get the new and Then we get the coarse-grained model of regional security dispatch, which is: Among them, FGW q，α (u, v) is the objective function of the coarse-grained model, α represents the trade-off parameter between node characteristics and node structure, q represents the distance parameter, and a i and b j are the node feature information of the equipment architecture and the support architecture, G1 is the equipment architecture, G2 is the support architecture, T i，j is the supply and demand resource matching between node i and node j, T k，l is the supply and demand resource matching between node k and node l, d * (a i , b j ) is α i and b j The characteristic distance between the distances between structures is integrated, is the result of integrating the distance between node i and node k in the two architectures into the distance within the structure, is the distance between node j and node l in the two architectures after integrating the distance within the structure, u i and v j They are the node data of the regional equipment architecture and the regional support architecture respectively.

4. The method for generating a regional equipment support scheduling plan based on structured data according to claim 2, characterized in that: Step S5 includes the following steps: S51: construct the fine-grained adjacency matrix of the regional equipment architecture and regional support architecture established in step S1 and S52, to the fine-grained adjacency matrix obtained in step S51 and Enter the node data x of the regional equipment architecture and regional support architecture respectively ik 、h ik and jl , g jl , normalize the node data h ik and g jl , and obtain the probability vector ph of each supply and demand point i and pg j ; S53, combining the regional security scheduling fine-grained model obtained in step S4 with the probability vector ph obtained in step S52 i and pg j Calculate X i To Y j The distribution distance And the guaranteed scheduling probability scheme PT ij ; S54, constructing coarse-grained adjacency matrices C1 and C2 of the regional equipment architecture and regional support architecture established in step S1; S55, input the node data x of the regional equipment system structure and the regional support system structure into the coarse-grained adjacency matrices C1 and C2 obtained in step S54 respectively. i 、u i and j 、v j , normalize the node data u i and v j , get the probability vectors pu and pv; S56: Calculate X based on the regional security scheduling coarse-grained model obtained in step S3 and the probability vectors pu and pv obtained in step S55. ik To Y jl The distribution distance FGW q,α (u,v) and the guaranteed scheduling probability scheme PT; S57: The guaranteed scheduling probability scheme PT obtained in step S53 ij The fine-grained model guarantees the scheduling solution T ij ; S58. Calculate the coarse-grained model guarantee scheduling scheme T according to the guarantee scheduling probability scheme PT obtained in step S56.

5. The method for generating a regional equipment support scheduling plan based on structured data according to claim 4 is characterized in that: Step S57 includes the following steps: S571: According to the guaranteed scheduling probability scheme PT obtained in step S53 ij , get the minimum resource allocation solution And the most resource delivery solution S572: Based on step S571, if sum(h i )≥sum(g j ), at this time, the demand can be met, and the fine-grained model guarantees the scheduling solution T ij Minimum resource allocation solution S573. Based on step S571, if sum(h i )<sum(g j ), at this time, the demand is partially met, then the fine-grained model guarantees the scheduling solution T ij Contains the most resource delivery solutions and scheduling scheme Two parts.

6. The method for generating a regional equipment support scheduling plan based on structured data according to claim 4 is characterized in that: Step S58 includes the following steps: S581, according to the guaranteed scheduling probability scheme PT obtained in step S56, obtain the minimum resource allocation scheme T 1 =PT*sum(v) and the maximum resource delivery plan T 2 =PT*sum(u); S582: Based on step S581, if sum(u)≥sum(v), the demand can be met, and the coarse-grained model guarantees that the scheduling solution T is the minimum resource configuration solution T. 1 ; S583: Based on step S581, if sum(u)<sum(v), the demand is partially met, and the coarse-grained model guarantees that the scheduling solution T contains the optimal guarantee solution and contains the most resource delivery solutions T. 2 and scheduling scheme sum(T 1 -T 2 , 2).

Images