Investment time sequence optimization method and system for technical improvement task of power grid

Abstract

The invention discloses an investment time sequence optimization method and system for a power grid technical improvement task, and relates to the technical field of power grid technical improvement. Splitting the target technical improvement task to obtain an initial relation graph; generating an initial triple set according to the initial relation graph, and obtaining a target triple sorting set according to the initial triple set and the initial relation graph; and generating a technical improvement investment time sequence set according to the initial relation graph and the target triple sorting set, screening the technical improvement investment time sequence set to obtain a target technical improvement investment scheme, and performing technical improvement according to the target technical improvement investment scheme. A target technical improvement task is split to generate an initial relation graph, an initial triple set is generated according to the initial relation graph to carry out relation calculation on different nodes in the initial relation graph, the priority sequence and the dependency relation of all tasks are defined, and then according to the initial relation graph and a target triple sorting set, the whole technical improvement scheme has the advantages that the technical improvement efficiency is improved. And under the condition of ensuring reasonable allocation of time, the investment cost is reduced, and the long-term economic benefit is improved.

Description

Technical Field

[0001] The invention relates to an investment timing optimization method and system for power grid technical transformation, belonging to the technical field of power grid technical transformation. Background Art

[0002] Grid technical transformation refers to the process of technical transformation of grid equipment or systems. Its purpose is to improve the operating efficiency, reliability, safety and economy of the grid by upgrading, optimizing or replacing existing equipment and technologies to adapt to new operating needs or solve existing operating problems.

[0003] Patent CN112668249B discloses an online construction method and system for overhaul and technical renovation plans for primary power grid equipment. The annual operation and maintenance data of primary power grid equipment is collected online. The data range covers the operation and maintenance records of equipment of the same type and manufacturer in different operating years. Based on this, a cost prediction model suitable for limited data scenarios is constructed, and the prediction model is used to evaluate the future operation and maintenance costs of the equipment. The economic life of the equipment is determined with the goal of minimizing the average annual cost of the equipment. Combined with the relationship between the current operating life of the equipment, the calculated economic life and the technical life of the equipment, a comprehensive analysis is performed to determine the optimal plan for whether the equipment needs technical transformation or overhaul. Finally, the generated technical transformation or overhaul plan is displayed in a visual form to provide intuitive decision support for equipment operation and maintenance and management.

[0004] However, existing power grid technical transformation plans generally have certain limitations, especially in terms of cost planning. Usually, technical transformation tasks are completed only based on the operating life of the equipment and basic economic requirements, while ignoring the priority order and dependencies between tasks. As a result, it is impossible to reasonably allocate tasks in time, and thus it is impossible to further optimize costs and improve benefits. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention provides an investment timing optimization method for power grid technical transformation tasks taking into account the priority order and dependency relationship between the tasks.

[0006] In order to achieve the above-mentioned purpose, the technical solution proposed by the present invention is: a method for optimizing the investment timing of power grid technical transformation tasks, comprising:

[0007] Acquire the attribute characteristics of the target technical transformation task of the power grid, split the target technical transformation task into a sub-technical transformation task set according to the attribute characteristics, and obtain an initial relationship graph according to the sub-technical transformation task set; the nodes in the initial relationship graph represent the sub-technical transformation tasks, and the edges represent the association relationship between the sub-technical transformation tasks;

[0008] Generate an initial triple set according to the initial relationship graph, and obtain a target triple sorted set according to the initial triple set and the initial relationship graph;

[0009] Generate a technological transformation investment sequence set according to the initial relationship diagram and the target triplet sorting set, calculate the economic benefits of each technological transformation investment sequence in the technological transformation investment sequence set and obtain a target score, and take the technological transformation investment sequence with the highest target score as the target technological transformation investment sequence plan.

[0010] A further design of the above technical solution is as follows: the attribute characteristics are project voltage level, project type, project nature, project distribution area and construction period.

[0011] Generating an initial triple set according to the initial relationship graph includes:

[0012] For the initial relationship graph, obtain the attribute characteristics of each node and the relationship type of each edge;

[0013] For each node, a portrait list is generated according to the attribute characteristics of the node, each node and edge in the graph is traversed, modeled in triple format, and the portrait list of each node is mapped to the triple to obtain the initial triple set.

[0014] The relationship types include must-build, coexistence, subordination and mutual exclusion

[0015] Obtaining a target triplet sorting set according to the initial triplet set and the initial relationship graph includes:

[0016] Step 1.1: taking the portrait list corresponding to each node in the initial relationship graph as the initial feature set;

[0017] Step 1.2: for the initial relationship graph, according to the target node, obtain the nodes adjacent to the target node to obtain a neighbor node set; the target node is any one in the initial relationship graph;

[0018] Step 1.3: Classify the neighbor nodes in the neighbor node set according to the relationship type to obtain first local neighbor node sets of different categories;

[0019] Step 1.4: Determine the local neighbor feature corresponding to the target node in each first local neighbor node set, and obtain the local neighbor features corresponding to all first local neighbor node sets to obtain a local neighbor feature set;

[0020] Step 1.5: updating the initial feature set of the target node according to all local neighbor feature sets, and using it as the target feature vector corresponding to the target node;

[0021] Step 1.6: Generate a test triple set according to the target feature vector corresponding to each node in the initial relationship graph;

[0022] Step 1.7: Taking the initial triple set as a comparison, for the test triple set, the Distmult decoder is used to calculate the score of each test triple, and all test triples are sorted according to the score to obtain the target triple sorting set.

[0023] Generating a technical transformation investment time series set according to the initial relationship graph and the target triplet sorting set includes:

[0024] Step 2.1: Determine the predecessor task and dependency table of each node according to the initial relationship graph and relationship type, and initialize the time series investment set and task completion set;

[0025] Step 2.2: Divide the total time period according to the preset time period to obtain a set of time segments in the time series, and initialize the available resources for each time period;

[0026] Step 2.3: Traverse all executable tasks within the target time period to obtain an executable task set, and select tasks in the executable task set in turn according to the target triplet sorting set to obtain an assignable task set; the executable task is a task whose predecessor tasks have all been completed; the target time period is any one of the time segments;

[0027] Step 2.4: If the target task satisfies the dependency relationship and has not been assigned, check whether there are resources in the current time period that satisfy the assignment of the target task; the target task is any one of the assignable tasks;

[0028] Step 2.5: If there are resources that meet the target task allocation, add the node corresponding to the target task to the current time period, move the target task from the assignable task set to the task completion set, update the dependency table, unlock the subsequent tasks that depend on the target task, and return to step 2.3;

[0029] Step 2.6: If there are resources that do not meet the target task allocation, the target task is postponed to the next time period, and the next task corresponding to the target task in the allocatable task set is updated as the target task, and step 2.3 is returned until the target task can be allocated or the allocatable task set is traversed;

[0030] Step 2.7: Record the next time period of the current time period as the target time period, and return to step 2.3 until all executable tasks in all time segments are traversed to obtain multiple technical transformation investment time series recorded as a technical transformation investment time series set.

[0031] The calculation method of the target score of the technical transformation investment time series is as follows:

[0032] The objective function is the benefit minus the cost of technical transformation investment sequence, and the budget ceiling, resource allocation restrictions and construction period requirements are used as constraints.

[0033] Calculate the number of sub-technical transformation tasks that do not meet the constraint conditions in the technical transformation investment time series, and use this as the value of the constraint condition;

[0034] The target score is obtained by subtracting the values ​​of all constraints from the value of the objective function.

[0035] A system for optimizing the investment sequence of power grid technical transformation tasks, the system comprising:

[0036] An initial relationship graph generation module is used to split the target technical transformation task into a sub-technical transformation task set by attribute characteristics, and obtain an initial relationship graph according to the sub-technical transformation task set; the nodes in the initial relationship graph represent the sub-technical transformation tasks, and the edges represent the association relationship between the sub-technical transformation tasks;

[0037] A target triplet sorting set generating module, used to generate an initial triplet set according to the initial relationship graph, and obtain a target triplet sorting set according to the initial triplet set and the initial relationship graph;

[0038] The target technological transformation investment plan determination module is used to generate a technological transformation investment timing set according to the initial relationship diagram and the target triplet sorting set, calculate the target score for the technological transformation investment timing set, and take the technological transformation investment timing with the highest target score as the target technological transformation investment timing plan.

[0039] Calculate the economic benefits of each technological transformation investment sequence in the technological transformation investment sequence set and obtain the target score. Take the technological transformation investment sequence with the highest target score as the target technological transformation investment sequence plan.

[0040] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, enable the computing device to execute the above-mentioned investment timing optimization method for power grid technical transformation tasks.

[0041] A computing device comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing the above-mentioned investment timing optimization method for power grid technical transformation tasks.

[0042] The beneficial effects of the present invention are:

[0043] The present invention proposes an investment timing optimization method for power grid technical transformation, which splits a target technical transformation task into a sub-technical transformation task set through attribute characteristics, and obtains an initial relationship graph according to the sub-technical transformation task set; the nodes in the initial relationship graph represent the sub-technical transformation tasks, and the edges represent the associations between the sub-technical transformation tasks; an initial triple set is generated according to the initial relationship graph, and a target triple sorting set is obtained according to the initial triple set and the initial relationship graph; a technical transformation investment timing set is generated according to the initial relationship graph and the target triple sorting set, the technical transformation investment timing set is screened to obtain a target technical transformation investment plan, and technical transformation is carried out according to the target technical transformation investment plan. By analyzing the attribute characteristics of the target technological transformation task, it is split into multiple sub-technical transformation tasks to generate an initial relationship graph, and then an initial triple set is generated based on the initial relationship graph. The relationship between different nodes in the initial relationship graph is calculated to obtain the target triple sorting set, which clarifies the priority order and dependency relationship of each task. Then, based on the initial relationship graph and the target triple sorting set, a technological transformation investment timing set is generated, so that the entire technological transformation plan can be executed in the optimal order, thereby improving investment efficiency. The entire technological transformation plan can reduce investment costs and improve long-term economic benefits while ensuring reasonable time allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A flow chart of a method for optimizing investment timing for power grid technical transformation is provided for an embodiment of the present invention;

[0045] Figure 2 A framework diagram of an investment timing optimization system for power grid technical transformation is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] Example

[0048] This embodiment provides a method for optimizing the investment timing for power grid technical transformation. Figure 1 As shown, it is a flowchart of a method for optimizing investment timing for power grid technical transformation provided by an embodiment of the present invention, comprising the following steps:

[0049] S1, split the target technical transformation task into a sub-technical transformation task set by attribute characteristics, and obtain the initial relationship graph based on the sub-technical transformation task set; the nodes in the initial relationship graph represent the sub-technical transformation tasks, and the edges represent the relationship between the sub-technical transformation tasks; it helps project managers understand the order, interdependence and possible conflicts between tasks, so as to make more reasonable arrangements and optimizations

[0050] The target technical transformation task is a task that requires technical transformation of the power grid. By splitting the target technical transformation task into multiple subtasks through attribute characteristics, the specific content, goals and requirements of each subtask can be more clearly identified, so that complex technical transformation tasks can be broken down into small manageable parts, reducing the complexity and uncertainty in the execution process.

[0051] In this embodiment, the attribute characteristics may be project voltage level, project type, project nature, project distribution area, construction period, etc.; voltage level refers to the working voltage of equipment in the power system, such as substations and transmission lines, for example, 35kV, 110kV, 220kV, 500kV and other different voltage levels; project type refers to the specific category or nature of the project, such as power projects, communication projects, environmental protection projects, etc.; project nature reflects the different goals and priorities of the project, such as "upgrade and transformation" projects, "new construction" projects or "expansion" projects; project distribution area refers to the geographical location of the project implementation, which may include urban and rural areas, different provinces and cities, etc.; construction period refers to the time required from the start of the project to the completion of the project. Association relationships include must-build, coexistence, subordination and mutual exclusion.

[0052] S2, generating an initial triple set according to the initial relationship graph, and obtaining a target triple sorting set according to the initial triple set and the initial relationship graph;

[0053] Generating an initial triple set according to the initial relationship graph includes:

[0054] For the initial relationship graph, obtain the attribute characteristics of each node and the relationship type of each edge;

[0055] For each node, a portrait list is generated based on the attribute characteristics of the node, each node and edge in the graph is traversed, modeled in triple format, and the portrait list of each node is mapped to the triple to obtain the initial triple set. The generated portrait includes voltage level, project type, project nature, distribution area, construction period, investment amount, etc. The portrait list is a collection of attribute characteristics of each node, which is used to describe various information and characteristics of the task.

[0056] The information in the initial relationship graph is standardized in the form of triples, so that the data structure is clear and easy to store and transmit; the triple format modeling is represented as (head entity, relationship type, tail entity), the head entity and the tail entity are sub-technical transformation tasks, and the head entity and the tail entity also include the corresponding portrait list; the relationship types include must-build, coexistence relationship, subordinate relationship and mutually exclusive relationship.

[0057] The target triplet sorting set obtained according to the initial triplet set and the initial relationship graph includes:

[0058] Step 1: Obtain the initial feature set according to the portrait list corresponding to each node in the initial relationship graph, that is, the portrait list corresponding to each node;

[0059] Step 2: For the initial relationship graph, according to the target node, obtain the i-th layer node (a total of N layers) adjacent to the target node to obtain the neighbor node set; the target node is any one in the initial relationship graph;

[0060] Step 3: Classify the neighbor nodes in the neighbor node set according to the relationship type to obtain the first local neighbor node sets of different categories;

[0061] Step 4: Determine the local neighbor features corresponding to the target node in each first local neighbor node set, and obtain the local neighbor features corresponding to all first local neighbor node sets to obtain a local neighbor feature set;

[0062] Local neighbor features refer to the feature information of nodes adjacent to the target node (i.e., neighbor nodes) in the graph, which represent the relationship and features between the target node and its direct neighbors (or multi-layer neighbors).

[0063] Determine the local neighbor features corresponding to the target node in each first local neighbor node set, specifically, substitute the graph formed by the current target node in each first local neighbor node set into the R-GCN network encoder. The own characteristics of each node are obtained by unique-hot encoding of its own project portrait, where must be built is recorded as 0, coexistence relationship is recorded as 1, subordination relationship is recorded as 2, and mutual exclusion relationship is recorded as 3. The transmission project is coded as: 1, 0, 0, the distribution project is coded as: 0, 1, 0, the substation project is coded as: 0, 0, 1, etc. The specific coding is determined by the technical staff.

[0064] Step 5: Update the features of the target node according to all local neighbor feature sets;

[0065] Step 6: If i is less than N, then i is incremented by 1 and the process returns to step 2 until i is equal to N. The target feature vector corresponding to the target node is obtained, which is the feature finally obtained by the above loop process.

[0066] Step 7: Generate a test triple set based on the target feature vector corresponding to each node in the initial relationship graph;

[0067] Step 8: Using the initial triple set as a comparison, for the test triple set, the Distmult decoder is used to calculate the score of each test triple, and all test triplets are sorted according to the score to obtain the target triple sorting set.

[0068] The construction of the initial feature set and the local neighbor feature set ensures that the features of the target node are not only derived from its own attributes, but also enhanced through the relationship features of multiple layers of neighbor nodes. The feature extraction and update of neighbor nodes of different categories make the feature vector of the target node more hierarchical and global, which can more accurately describe the dependency and impact range of the task. The layer-by-layer neighbor feature extraction can capture the global information within the multi-hop range, not only focusing on directly related tasks, but also considering the impact of indirectly related tasks, to achieve global feature aggregation.

[0069] Neighbor nodes are classified according to the relationship type, decomposing the originally complex network relationships into a finer-grained semantic hierarchy. Each set of neighbor nodes describes a certain type of association characteristics of the target node, such as must-build, coexistence, subordination, and mutual exclusion. This enhances the semantic expression of the relationships between technical transformation tasks, can more clearly identify key tasks and secondary tasks, and optimize resource allocation.

[0070] The target feature vector generated by each target node is an encoded expression of comprehensive information such as itself, neighboring nodes, and relationship types. It can comprehensively and accurately characterize the importance of the task. The generation of feature vectors and the subsequent triplet scoring process can scientifically sort the priorities of tasks and provide a basis for technological transformation investment and resource allocation.

[0071] In this embodiment, the feature update of the target node according to all local neighbor feature sets is specifically performed by formula

[0072]

[0073] in, is the initial feature of the target node o; R is the relationship type, including {must be established, coexistence relationship, subordination relationship, mutual exclusion relationship}; p is the neighbor node of the target node; is the set of neighbor nodes with a relationship r with the target node; the value range of r is {0,1,2,3}; C is The absolute value of represents the number of neighbor nodes with a relationship r with the target node; is the initial feature of the neighbor node P corresponding to the target node at the i-th layer; The weight matrix formed by the neighbor nodes with a relationship of r relative to the target node at the i-th layer is specifically a weight matrix obtained by constructing a matrix with values ​​of 0 based on the target node and all neighboring nodes. According to the relationship r, when the relationship between the node and the target node is r, the corresponding value in the matrix is ​​modified to 1; α is the sigmoid activation function.

[0074] For the target node o, determine the set of neighbor nodes at the i-th layer from the relationship graph, and classify the neighbor nodes according to the relationship type r to obtain the neighbor nodes under different categories. Each relationship type corresponds to a set of neighbor nodes; for each relationship r, through the weight matrix The neighbor node features are weighted to obtain aggregate features. After all the aggregate results are normalized, the aggregate results of neighbor features of all relationship types are added together to obtain the overall features of the neighbor nodes. The overall features are added to the initial features, and the updated features of the target node are generated through the activation function sigmoid.

[0075] In this embodiment, the score of each test triple is calculated by the Distmult decoder. Specifically, for each triple (s, r, o), the eigenvector corresponding to node s is us, and the eigenvector corresponding to node o is uo. The dimension d of the eigenvector is obtained, and a d×d diagonal matrix Wr is generated. For each test triple, the product of the diagonal matrix and the eigenvector uo is calculated to obtain the relationship matrix, and the transpose of the eigenvector us is multiplied by the relationship matrix to obtain the target scalar. The target scalar is applied with the Sigmoid activation function to obtain the score.

[0076] By scoring the test triples through the DistMult decoder, it is possible to verify whether the generated feature vectors and triples are consistent with the original data, so as to find possible errors or optimization points; according to the sorted set, it is possible to clarify which tasks have priority investment and which tasks can be postponed, so as to maximize the benefits of technical transformation investment, avoid low-priority tasks from occupying too many resources, and ensure the efficient use of funds, time and human resources; the layer nodes adjacent to the target node are specifically judged as follows: if there is a path from the node to the target node where there is no node, then the node is recorded as the adjacent node of the target node in the first layer; if the path from the node to the target node must pass through at least 2 nodes, then the node is recorded as the adjacent node of the target node in the third layer.

[0077] S103, generating a technological transformation investment time series set according to the initial relationship graph and the target triplet sorting set, screening the technological transformation investment time series set to obtain a target technological transformation investment plan, and performing technological transformation according to the target technological transformation investment plan.

[0078] By generating a technical transformation investment time sequence set, the timing and sequence of implementation of each sub-task and technical transformation step can be clarified, which helps to reasonably arrange time and capital investment in the investment and avoid over-concentrated or over-dispersed investment arrangements.

[0079] In this embodiment, generating a technical transformation investment time series set according to the initial relationship graph and the target triplet sorting set includes:

[0080] Step 1: Determine the predecessor tasks and dependency table of each node according to the initial relationship graph, and initialize the timing investment set and task completion set;

[0081] Step 2: Divide the total time period according to the preset time period to obtain a set of time segments in the time series, and initialize the available resources for each time period;

[0082] Step 3: Traverse all executable tasks in the target time period to obtain an executable task set, and select tasks in the executable task set in turn according to the target triplet sorting set to obtain an assignable task set; an executable task is one whose predecessor tasks have all been completed; the target time period is any one of the time segments; if all predecessor tasks of a task have been completed, it is an executable task;

[0083] Step 4: If the target task satisfies the dependency and has not been assigned, check whether there are resources in the current time period that can satisfy the target task assignment; the target task is any one of the assignable tasks;

[0084] Step 5: If there are resources that meet the target task allocation, add the node corresponding to the target task to the current time period, move the target task from the allocatable task set to the task completion set, update the dependency table, unlock the subsequent tasks that depend on the target task, and return to step 3;

[0085] Step 6: If there are resources that do not meet the target task allocation, the target task is postponed to the next time period, and the next task corresponding to the target task in the allocatable task set is updated as the target task, and return to step 3 until the target task can be allocated or the allocatable task set is traversed;

[0086] Step 7: Record the next time period of the current time period as the target time period, and return to step 3 until all executable tasks in all time segments are traversed to obtain multiple technical transformation investment time series recorded as a technical transformation investment time series set.

[0087] By clarifying the predecessor tasks of each task, the logic and correctness of the task chain are ensured, conflicts and logical confusions between tasks are avoided, and a clear implementation path is provided for the technical transformation investment plan. Subsequent tasks can only be executed after the predecessor tasks are completed. Dynamic allocation is carried out in each time period based on available resources and task requirements to avoid waste or shortage of resources. When resources are insufficient, tasks are postponed to the next time period, effectively balancing the priority of task execution and resource utilization.

[0088] The preset time period is determined by the technician; the available resources include the upper limit of funds and the upper limit of the number of tasks; the satisfied dependencies are represented as executable tasks

[0089] The system can manage multiple technical transformation tasks simultaneously, and ensure the coordinated advancement of all tasks through sharding and dependency updates; divide the time period of the entire technical transformation project into multiple stages through time segmentation to facilitate task scheduling and progress monitoring; dynamically update task status, adjust task allocation and execution plans at any time, and ensure flexible response to changes in project progress.

[0090] Based on the sorted set of target triples, high-priority tasks are assigned first to ensure that key tasks are completed first and improve the benefits of technical transformation; the problem of low-priority tasks occupying key resources and causing delays in the overall project progress is avoided; after each task is completed, the subsequent tasks it depends on are automatically unlocked, allowing the subsequent tasks to quickly enter the execution state, ensuring that resources can be seamlessly transferred to new tasks.

[0091] In response to emergencies during task execution, the system can dynamically adjust task arrangements to ensure optimal execution of the overall project; through the technical transformation investment time sequence set, the investment focus of each time period can be clearly planned to avoid excessive dispersion or idleness of funds, and the investment plan can be dynamically adjusted according to the actual execution progress to improve fund utilization.

[0092] In this embodiment, the target technological transformation investment plan obtained by screening the technological transformation investment time series set includes:

[0093] For each technological transformation investment sequence in the technological transformation investment sequence set, the economic benefit of the technological transformation investment sequence is calculated to obtain the target score;

[0094] The technological transformation investment sequence with the highest target score is recorded as the target technological transformation investment plan.

[0095] Taking economic benefits as the core evaluation indicator, by calculating the target score of each technological transformation investment sequence, the most efficient plan is selected to ensure the return on investment and maximize the value of the technological transformation investment, avoid blind investment, give priority to the implementation of technological transformation plans with significant benefits, and achieve optimal allocation of resources.

[0096] Calculate the economic benefits of the technical transformation investment sequence to get the target score. Specifically, for each technical transformation investment sequence, combine the objective function and the penalty function to calculate the objective function. The technical transformation investment sequence includes the specific task execution sequence and time period resource allocation. The economic objective function is a mathematical expression for calculating the economic value based on investment return, operating cost reduction, and benefit growth. The constraints are the budget cap, resource allocation restrictions, and construction period requirements, which are denoted as g(x), h(x), and e(x). Where i represents a sub-technical transformation task, and N represents the number of sub-technical transformation tasks. The objective function is fe(x), which is the benefit minus the cost. The benefit is determined by the technical personnel after completion, and the cost includes the investment amount, operation and maintenance cost, and resource use cost. For the technical transformation investment sequence, check whether it violates the constraints g(x), h(x), and e(x). If the constraints are not violated, g(x), h(x), and e(x) are all zero. If there is a violation of the constraints, determine how many sub-technical transformation tasks violate the constraints. For example, if there are 2 g(x), then g(x) takes a value of 4. The target score is obtained by subtracting the values ​​of all constraints from the value of the objective function.

[0098] This embodiment analyzes the attribute characteristics of the target technological transformation task and splits it into multiple sub-technical transformation tasks, thereby generating an initial relationship graph, and then generating an initial triple set based on the initial relationship graph, performing relationship calculations on different nodes in the initial relationship graph to obtain a target triple sorting set, clarifying the priority order and dependency relationship of each task, and then generating a technological transformation investment timing set based on the initial relationship graph and the target triple sorting set, so that the entire technological transformation plan can be executed in the optimal order, thereby improving investment efficiency, so that the entire technological transformation plan, while ensuring reasonable time allocation, also reduces investment costs and improves long-term economic benefits.

[0099] Embodiment 2

[0100] This embodiment provides an investment timing optimization system for power grid technical transformation. Figure 2 A framework diagram of an investment timing optimization system for power grid technical transformation provided by an embodiment of the present invention includes:

[0101] An initial relationship graph generation module is used to split the target technical transformation task into a sub-technical transformation task set by attribute characteristics, and obtain an initial relationship graph according to the sub-technical transformation task set; the nodes in the initial relationship graph represent the sub-technical transformation tasks, and the edges represent the associations between the sub-technical transformation tasks;

[0102] A target triplet sorting set generation module is used to generate an initial triplet set according to an initial relationship graph, and obtain a target triplet sorting set according to the initial triplet set and the initial relationship graph;

[0103] The target technological transformation investment plan determination module is used to generate a technological transformation investment time series set according to the initial relationship graph and the target triplet sorting set, screen the technological transformation investment time series set to obtain the target technological transformation investment plan, and carry out technological transformation according to the target technological transformation investment plan.

[0104] An investment timing optimization system for power grid technical transformation provided by an embodiment of the present invention analyzes the attribute characteristics of a target technical transformation task and splits it into multiple sub-technical transformation tasks, thereby generating an initial relationship graph, and then generating an initial triple set based on the initial relationship graph, performing relationship calculations on different nodes in the initial relationship graph to obtain a target triple sorting set, clarifying the priority order and dependency relationship of each task, and then generating a technical transformation investment timing set based on the initial relationship graph and the target triple sorting set, so that the entire technical transformation plan can be executed in the optimal order, thereby improving investment efficiency, so that the entire technical transformation plan, while ensuring reasonable time allocation, also reduces investment costs and improves long-term economic benefits.

[0105] In one embodiment, the target triplet sorting set generation module includes:

[0106] The initial relationship graph feature determination module is used to obtain the attribute features of each node and the relationship type of each edge of the initial relationship graph;

[0107] The triple determination module is used to generate a portrait list for each node according to the attribute characteristics of the node, traverse each node and edge in the graph, model it in triple format, and map the portrait list of each node into the triple to obtain the initial triple set.

[0108] In one embodiment, the target triplet sorting set generation module further includes:

[0109] An initial feature set determination module is used to obtain an initial feature set according to a portrait list corresponding to each node in the initial relationship graph;

[0110] A neighbor node determination module is used to obtain the i-th layer node adjacent to the target node according to the initial relationship graph to obtain a neighbor node set; the target node is any one in the initial relationship graph;

[0111] A node classification module, used for classifying neighbor nodes in the neighbor node set according to the relationship type to obtain first local neighbor node sets of different categories;

[0112] A local neighbor feature set determination module is used to determine the local neighbor features corresponding to the target node in each first local neighbor node set, and obtain the local neighbor features corresponding to all first local neighbor node sets to obtain a local neighbor feature set;

[0113] A feature updating module is used to update the features of the target node according to all local neighbor feature sets;

[0114] The target feature vector generation module is used to increase i by one if i is less than N, and return to the neighbor node determination module until i is equal to N, so as to obtain the target feature vector corresponding to the target node;

[0115] A test triple determination module is used to generate a test triple set according to a target feature vector corresponding to each node in the initial relationship graph;

[0116] The test triplet sorting module is used to compare the initial triplet set, calculate the score of each test triplet for the test triplet set through the Distmult decoder, and sort all the test triples according to the score to obtain the target triplet sorting set.

[0117] In one embodiment, the target technological transformation investment plan determination module includes:

[0118] An initialization module is used to determine the predecessor task and dependency table of each node according to the initial relationship graph, and initialize the timing investment set and task completion set;

[0119] The time segmentation module is used to segment the total time period according to the preset time period to obtain a set of time segments in the time series and initialize the available resources for each time period;

[0120] The assignable task set determination module is used to traverse all executable tasks in the target time period to obtain an executable task set, and select tasks in the executable task set in turn according to the target triplet sorting set to obtain an assignable task set; the executable task is that all the predecessor tasks of the execution task have been completed; the target time period is any one of the time segment set;

[0121] The resource allocation judgment module is used to check whether there are resources in the current time period that meet the target task allocation if the target task meets the dependency relationship and has not been allocated; the target task is any one of the allocable tasks;

[0122] A dependency table update module is used to add the node corresponding to the target task to the current time period if there are resources that meet the target task allocation, and move the target task from the allocatable task set to the task completion set, update the dependency table, unlock the subsequent tasks that depend on the target task, and return to the allocatable task set determination module;

[0123] The task update module is used to postpone the target task to the next time period if there are resources that do not meet the target task allocation, and update the next task corresponding to the target task in the allocatable task set as the target task, and return to the allocatable task set determination module until the target task can be allocated or the allocatable task set is traversed;

[0124] The time segment updating module is used to record the next time segment of the current time segment as the target time segment, and return to the assignable task set determination module until all executable tasks in all time segments are traversed to obtain multiple technical transformation investment timings recorded as a technical transformation investment timing set.

[0125] In one embodiment, the target technological transformation investment plan determination module further includes:

[0126] A target score calculation module, for calculating the economic benefits of each technological transformation investment sequence in the technological transformation investment sequence set to obtain a target score;

[0127] The technical transformation investment sequence screening module is used to obtain the technical transformation investment sequence with the highest target score as the target technical transformation investment plan.

[0128] Embodiment 3

[0129] This embodiment provides a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a computing device, the computing device executes the investment timing optimization method for power grid technical transformation tasks as described in Example 1.

[0130] This embodiment also provides a computing device, characterized in that it includes: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing the investment timing optimization method for power grid technical transformation tasks as described in Example 1.

[0131] The technical solutions of the present invention are not limited to the above-mentioned embodiments, and any technical solutions obtained by equivalent replacement methods fall within the scope of protection required by the present invention.

Claims

1. A method for optimizing the investment timing of power grid technical transformation tasks, characterized in that: include: Acquire the attribute characteristics of the target technical transformation task of the power grid, split the target technical transformation task into a sub-technical transformation task set according to the attribute characteristics, and obtain an initial relationship diagram according to the sub-technical transformation task set; The nodes in the initial relationship graph represent sub-technical transformation tasks, and the edges represent the associations between the sub-technical transformation tasks; Generate an initial triple set according to the initial relationship graph, and obtain a target triple sorted set according to the initial triple set and the initial relationship graph; Generate a technological transformation investment sequence set according to the initial relationship diagram and the target triplet sorting set, calculate the economic benefits of each technological transformation investment sequence in the technological transformation investment sequence set and obtain a target score, and take the technological transformation investment sequence with the highest target score as the target technological transformation investment sequence plan.

2. The method for optimizing the investment timing for power grid technical transformation tasks according to claim 1 is characterized in that: The attribute characteristics are project voltage level, project type, project nature, project distribution area and construction period.

3. The method for optimizing investment timing for power grid technical transformation tasks according to claim 2 is characterized in that: Generating an initial triple set according to the initial relationship graph includes: For the initial relationship graph, obtain the attribute characteristics of each node and the relationship type of each edge; For each node, a portrait list is generated according to the attribute characteristics of the node, each node and edge in the graph is traversed, modeled in triple format, and the portrait list of each node is mapped to the triple to obtain the initial triple set.

4. The method for optimizing investment timing for power grid technical transformation tasks according to claim 3 is characterized in that: The relationship types include must-build, coexistence, subordination and mutual exclusion.

5. The method for optimizing investment timing for power grid technical transformation tasks according to claim 4 is characterized in that: Obtaining a target triplet sorting set according to the initial triplet set and the initial relationship graph includes: Step 1.1: taking the portrait list corresponding to each node in the initial relationship graph as the initial feature set; Step 1.2: for the initial relationship graph, according to the target node, obtain the nodes adjacent to the target node to obtain a neighbor node set; the target node is any one in the initial relationship graph; Step 1.3: Classify the neighbor nodes in the neighbor node set according to the relationship type to obtain first local neighbor node sets of different categories; Step 1.4: Determine the local neighbor feature corresponding to the target node in each first local neighbor node set, and obtain the local neighbor features corresponding to all first local neighbor node sets to obtain a local neighbor feature set; Step 1.5: updating the initial feature set of the target node according to all local neighbor feature sets, and using it as the target feature vector corresponding to the target node; Step 1.6: Generate a test triple set according to the target feature vector corresponding to each node in the initial relationship graph; Step 1.7: Taking the initial triple set as a comparison, for the test triple set, the Distmult decoder is used to calculate the score of each test triple, and all test triples are sorted according to the score to obtain the target triple sorting set.

6. The method for optimizing investment timing for power grid technical transformation tasks according to claim 5 is characterized in that: Generating a technical transformation investment time series set according to the initial relationship graph and the target triplet sorting set includes: Step 2.1: Determine the predecessor task and dependency table of each node according to the initial relationship graph and relationship type, and initialize the time investment set and task completion set; Step 2.2: Divide the total time period according to the preset time period to obtain a set of time segments in the time series, and initialize the available resources for each time period; Step 2.3: Traverse all executable tasks within the target time period to obtain an executable task set, and select tasks in the executable task set in turn according to the target triplet sorting set to obtain an assignable task set; the executable task is a task whose predecessor tasks have all been completed; the target time period is any one of the time segments; Step 2.4: If the target task satisfies the dependency relationship and has not been assigned, check whether there are resources in the current time period that satisfy the assignment of the target task; the target task is any one of the assignable tasks; Step 2.5: If there are resources that meet the target task allocation, add the node corresponding to the target task to the current time period, move the target task from the assignable task set to the task completion set, update the dependency table, unlock the subsequent tasks that depend on the target task, and return to step 2.3; Step 2.6: If there are resources that do not meet the target task allocation, the target task is postponed to the next time period, and the next task corresponding to the target task in the allocatable task set is updated as the target task, and step 2.3 is returned until the target task can be allocated or the allocatable task set is traversed; Step 2.7: Record the next time period of the current time period as the target time period, and return to step 2.3 until all executable tasks in all time segments are traversed to obtain multiple technical transformation investment time series recorded as a technical transformation investment time series set.

7. The method for optimizing investment timing for power grid technical transformation tasks according to claim 6 is characterized in that: The calculation method of the target score of the technical transformation investment time series is as follows: The objective function is the benefit minus the cost of technical transformation investment sequence, and the budget ceiling, resource allocation restrictions and construction period requirements are used as constraints. Calculate the number of sub-technical transformation tasks that do not meet the constraint conditions in the technical transformation investment time series, and use this as the value of the constraint condition; The target score is obtained by subtracting the values ​​of all constraints from the value of the objective function.

8. An optimization system for the investment timing optimization method for power grid technical transformation tasks as described in claim 1, characterized in that: include: An initial relationship graph generation module is used to split the target technical transformation task into a sub-technical transformation task set by attribute characteristics, and obtain an initial relationship graph according to the sub-technical transformation task set; The nodes in the initial relationship graph represent sub-technical transformation tasks, and the edges represent the associations between the sub-technical transformation tasks; A target triplet sorting set generating module, used to generate an initial triplet set according to the initial relationship graph, and obtain a target triplet sorting set according to the initial triplet set and the initial relationship graph; The target technological transformation investment plan determination module is used to generate a technological transformation investment timing set according to the initial relationship diagram and the target triplet sorting set, calculate the target score for the technological transformation investment timing set, and take the technological transformation investment timing with the highest target score as the target technological transformation investment timing plan. Calculate the economic benefits of each technological transformation investment sequence in the technological transformation investment sequence set and obtain the target score, and take the technological transformation investment sequence with the highest target score as the target technological transformation investment sequence plan.

9. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, enable the computing device to execute the investment timing optimization method for power grid technical transformation tasks as described in any one of claims 1 to 7.

10. A computing device, characterized in that include: One or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing the investment timing optimization method for power grid technical transformation tasks as described in any one of claims 1 to 7.

Images