AI technology-based steel mill substation design scheme real-time optimization method and device
By applying AI technology in the design of steel mill substations, including data containers, greedy algorithms, dynamic programming and reinforcement learning models, the problems of high design pressure and data are easily missed, and the accuracy, real-time and efficient design solutions are achieved. .
Patent Information
- Application Number
- CN202411462300.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-18
- Publication Date
- 2025-02-11
AI Technical Summary
During the design process of steel plant substations, the design pressure is too high, data is prone to errors and missed, and the rationality of the plan is not reasonable, resulting in tight design cycles and easily causing data errors and misses.
The real-time optimization method of design scheme based on AI technology is adopted, and the total load of the substation is calculated through the data container, combined with a combination of greedy algorithm and dynamic programming algorithm for initial planning, and the intelligent body model with reinforcement learning is used to dynamically adjust the optimization layout plan.
The intelligent design of substation solution is realized, the accuracy, real-time and efficient design solution is ensured, and the design pressure and the risk of data errors and errors are reduced.
Smart Images
Figure CN119416950A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of steel plant substation design scheme, and in particular, relates to a real-time optimization method and device for a steel plant substation design scheme based on AI technology. Background Art
[0002] During the design phase of a steel plant project, the main professional will arrange the plant area according to the process system flow and the scope specified by the project. After the main workshop location is determined, the public and auxiliary professional will arrange the supporting facilities according to the remaining space in the plant area. However, steel plant projects have the characteristics of complex process flow, diverse process systems, and a wide variety of equipment. Therefore, during the design process, the public and auxiliary professional needs to adjust the plan in real time according to the design changes of the upstream professional, and modify the parameters and layout of the supporting facilities.
[0003] In the design process of the steel plant substation scheme, the electrical professionals will first make preliminary space occupation based on the number and location of the process systems and the remaining space in the plant. The initial plan is generally based on one process system supporting one substation. When the plant space is limited, the electrical professional designers need to adjust the plan according to the actual situation to ensure the normal power supply and distribution of the plant while ensuring the number and location of substations are reasonable. In addition, when the main professional design changes, the load of each distribution system will change due to the adjustment of parameters such as the number of equipment and power, and the size of the substation may also change accordingly. At the same time, the design of steel plant project schemes generally has the characteristics of short time cycle, large scheme changes, and limited site design. Therefore, the design scheme of steel plant substations often needs to be passively adjusted and the design cycle is tight.
[0004] In the design process of the substation scheme of the existing steel plant project, the designers first plan the number and location of the substations according to the open space conditions within the limited range of the plant area and the division of the process system. When the number and location of the substations are basically determined, the designers determine the number and size of the distribution cabinets and transformers in each substation according to the load calculation results, and finally determine the size of the substation. When the size, number and location of the substation are determined, the design of the substation scheme for the plant area is completed at this stage. However, when the number of equipment, power and other parameters in each process system change, the electrical professional designers need to restart the load calculation of the corresponding system in real time and adjust the size of the substation in real time according to the calculation results. When the plant layout changes, electrical designers need to re-determine the size, quantity and location of the substation based on the latest open space. If the plant space is limited, two or more substations need to be merged. The merged substation needs to distribute power to two or more systems. The previous substation load calculations will be restarted, and the initial design scheme will be completely overturned. At this time, designers need to complete the latest substation design scheme in a short period of time, and the scheme adjustment work will continue until the project plan is implemented. This process will generate a lot of data reorganization and calculation, and the design cycle is very tight, which is easy to cause data errors and omissions, causing great pressure on electrical professional design. Summary of the invention
[0005] The main purpose of the embodiments of the present invention is to provide a real-time optimization method and device for a steel plant substation design scheme based on AI technology, which solves the problems of excessive design pressure, easy data errors and omissions, and low rationality of the scheme during the substation scheme design process. By using AI technology, the intelligent design of the substation scheme is realized, effectively ensuring the accuracy, real-time and efficiency of the design scheme.
[0006] First, a real-time optimization method for a steel plant substation design scheme based on AI technology is provided, the method comprising: loading all data required for substation design into a data container, and applying the data container to calculate the total load of each substation; converting the CAD design drawing into an area placeholder map represented by grid blocks by layers, and performing initial substation planning based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm; taking the minimum distance between the overall substation and the process system as a reward target, using the initial substation planning as an initial state input into an intelligent agent model based on reinforcement learning, and dynamically adjusting the optimized layout plan based on updated data to obtain a final design plan.
[0007] In a possible implementation, using a data container to calculate the total load of a substation includes: using the data container to calculate the total load of electricity consumption of each process system by using the following relationship, and summing up to determine the total load of each substation:
[0008]
[0009] Among them, Sjs is the total apparent power of the substation, Pjs is the active power of each workshop, Qjs is the reactive power of each workshop, and m is the number of workshops powered by the substation.
[0010] In another possible implementation, the CAD design drawing is converted into an area placeholder map represented by grid blocks according to the layers, and the initial substation planning is performed based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm, including: the CAD design drawing is represented as a process system layer and a pipe network bridge layer according to the layers, and the location information and unoccupied area information of each process system in the current plant are obtained by converting the pipe network bridge layer and the process system layer into a grid block representation and storing them in a matrix; the unoccupied area information is divided into a set of usable empty areas and a pipe network occupied area according to the minimum area constraint of the boundary geometry; a path-finding algorithm is applied to form a set of shortest paths from each process system through the pipe network occupied area to each empty area; and the initial substation planning is performed based on the empty area set and the shortest path set using a combination of a greedy algorithm and a dynamic programming algorithm.
[0011] In another possible implementation, initial substation planning is performed based on the empty area set and the shortest path set using a combination of a greedy algorithm and a dynamic programming algorithm, including: if the number of available empty area sets is greater than the process system set data, the greedy algorithm is preferentially used for initial substation planning; if the number of available empty area sets is less than the process system set data, a dynamic programming strategy is used for initial substation empty area planning.
[0012] In another possible implementation, a greedy algorithm is used for initial substation planning, including: allocating according to the empty areas from large to small, giving priority to substation allocation for process systems with high load calculations; allocating a substation that meets the proximity principle to each process system in each empty space, starting with the empty area closest to the process system, and each allocation occupies the empty area of the substation according to the principle of closest distance.
[0013] In another possible implementation, a dynamic programming strategy is used for initial substation empty area planning, including: assuming that dp[i][j][k] is the cable consumption required to place the substation corresponding to the ith process system in the jth empty area without exceeding the empty area capacity k; determining the state transition process of substation empty area planning, including: if the substation of the ith process system is not placed in the jth empty area, then dp[i][j][k] = dp[i-1][j][k]; if the ith process system is placed in the jth empty area, then dp[i][j][k] = dp[i-1][j][k]. p[i-1][j][kw[i]]+v[i]), w[i] is the power supply load of the substation corresponding to the i-th process system, and v[i] is the cable consumption of the substation in the empty area; according to the state transfer process, the state transfer equation is obtained: dp[i][j][k]=min(dp[i-1][j][k],dp[i-1][j][kw[i]]+v[i]); according to the state transfer equation, the initial substation empty area planning corresponding to the minimum cable consumption is solved when the substations corresponding to all process systems are placed in all empty areas.
[0014] In another possible implementation, the distance between the overall substation and the process system is minimized as a reward target, the initial substation planning is used as an initial state input to the reinforcement learning-based intelligent model, and the layout plan is dynamically adjusted and optimized based on updated data to obtain a final design plan, including: based on the design data of historical engineering projects, the reinforcement learning-based intelligent model is trained with the minimum cable consumption as the goal to obtain a trained reinforcement learning-based intelligent model; the distance between the overall substation and the process system is minimized as a reward target, and the initial substation planning is used as an initial state input to the trained reinforcement learning-based intelligent model; based on the updated data, the data container is used to synchronize data in real time, the load calculation of the corresponding process system is adjusted, the substation demand is dynamically planned and updated, and the trained reinforcement learning-based intelligent model is used to iteratively adjust the design plan until the current project requirements are met to obtain a final design plan.
[0015] Secondly, a real-time optimization device for steel plant substation design scheme based on AI technology is provided, and the device includes: a data acquisition module, which is used to load all data required for substation design into a data container, and use the data container to calculate the total load of each substation; an initial planning module, which is used to convert the CAD design drawing into an area placeholder map represented by grid blocks by layer, and use a combination of a greedy algorithm and a dynamic programming algorithm to perform initial substation planning according to the area placeholder map; a dynamic adjustment module, which is used to minimize the distance between the overall substation and the process system as a reward target, use the initial substation planning as the initial state input to the intelligent model based on reinforcement learning, and dynamically adjust the optimization layout plan based on the updated planning parameters to obtain the final design plan.
[0016] In a third aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, a real-time optimization method for a steel plant substation design scheme based on AI technology as provided in the first aspect is implemented.
[0017] In a fourth aspect, a non-transitory computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, a real-time optimization method for a steel plant substation design scheme based on AI technology as provided in the first aspect is implemented. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.
[0019] Figure 1 A flowchart of a method for real-time optimization of a steel plant substation design scheme based on AI technology provided by an embodiment of the present invention;
[0020] Figure 2 A schematic diagram of building an intelligent agent model based on reinforcement learning according to an embodiment of the present invention;
[0021] Figure 3 A structural diagram of a real-time optimization device for a steel plant substation design scheme based on AI technology provided by an embodiment of the present invention;
[0022] Figure 4 A schematic diagram of the physical structure of an electronic device provided by the present invention.
[0023] Specific implementation method
[0024] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.
[0025] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, modules and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, modules, components and / or groups thereof. It should be understood that when we refer to a module as being "connected" or "coupled" to another module, it may be directly connected or coupled to the other modules, or there may be intermediate modules. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any modules and all combinations of one or more associated listed items.
[0026] Since steel plant projects have the characteristics of complex process flow, a wide variety of equipment, limited plant space, and large volume of changes, the following disadvantages are likely to occur when designing steel plant substation schemes: when a large amount of equipment information changes, it is necessary to recheck the equipment data, repeat the load calculation, and restart the substation design. A large amount of repeated work will greatly reduce work efficiency; when the layout of the plant is adjusted and the remaining space changes greatly compared to before, the layout and size of the substation need to be replanned, the utilization rate of the original calculation results is low, and the data reorganization process will also generate a large amount of repeated work; the time cycle of the steel plant project scheme design phase is short. When the change is large, the reserved design time of the electrical profession is tight and the design pressure is large. In view of the disadvantages of the existing steel plant substation design scheme, the present invention provides a real-time optimization method and device for the steel plant substation design scheme based on AI technology, which ensures the real-time reorganization and adjustment of the calculation data, and ensures that the calculation data is effectively attached to the designed substation. No matter how the main profession makes design changes, the electrical profession can quickly complete the latest scheme design in real time, thereby effectively ensuring the accuracy, real-time and efficiency of the design scheme.
[0027] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation method of the present application will be further described in detail below in conjunction with the accompanying drawings.
[0028] The technical solution of the present application and how to solve the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0029] like Figure 1 The figure is a flow chart of a method for real-time optimization of a steel plant substation design scheme based on AI technology provided by an embodiment of the present invention, the method comprising:
[0030] Step S11, loading all data required for substation design into a data container, and using the data container to calculate the total load of each substation.
[0031] First, all the data in the upstream professional commissioned materials are loaded into the data container, which contains the basic fields required for substation design, such as load calculation. Among them, the equipment name and the workshop (location) to which it belongs are combined as unique data record identifiers. The total power load of each process system is calculated through the data container, which helps to determine the total load of the substation, thereby determining the size and internal layout of the substation. The total load of the substation is the sum of the loads of all workshops powered by the substation.
[0032] Step S12, converting the CAD design drawing into an area placeholder map represented by grid blocks according to the layer, and performing initial substation planning based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm.
[0033] The CAD design drawings are represented by layers as process system layers and pipe network bridge layers. The CAD design drawings are converted into regional placeholder maps represented by grid blocks by layers to obtain the current plant area location information set and the remaining empty areas. Subsequently, the combination of greedy algorithm and dynamic programming algorithm is used for the initial substation planning.
[0034] Step S13, taking the minimum distance between the overall substation and the process system as the reward target, using the initial substation planning as the initial state input to the intelligent model based on reinforcement learning, dynamically adjusting the optimized layout plan based on the updated data, and obtaining the final design plan.
[0035] The embodiment of the present invention performs real-time optimization of the steel plant substation design scheme based on artificial intelligence (AI) technology, solves the problems of excessive design pressure, easy data errors and omissions, and low rationality of the scheme during the substation scheme design process, and realizes the intelligent design of the substation scheme by using AI technology, effectively ensuring the accuracy, real-time and efficiency of the design scheme.
[0036] In the embodiment of the present invention, in step S11, the total power load of each process system is calculated by applying the following relationship through the data container, and the total load of each substation is determined by summing:
[0037]
[0038] Among them, Sjs is the total apparent power of the substation, Pjs is the active power of each workshop, Qjs is the reactive power of each workshop, and m is the number of workshops powered by the substation. The data container carries the basic fields required for load calculation of all distribution equipment in the current m workshops.
[0039] In step S12, optionally, the CAD design drawing is first represented as a process system layer and a pipe network bridge layer according to the layer, and the pipe network bridge layer and the process system layer are converted into a grid block representation and stored in a matrix to obtain the occupied location information and unoccupied area information of each process system in the current plant. The occupied location information is represented as a set {S1, S2, S3....Sn}, and the unoccupied area information is the currently remaining empty area.
[0040] The unoccupied area information is divided into a usable empty area set and a pipe network occupied area according to the minimum area constraint of the boundary geometry. For example, an area greater than the constant square meter of the fixed constraint is determined as a usable empty area. The usable empty area set is recorded as {E1, E2, E3...Em}.
[0041] Then, a path-finding algorithm is applied to form a set of shortest paths {Lf1e1, Lf1e2, Lf1e3....Lf1em....Lfnem} from each process system through the occupied area of the pipeline network to each empty area.
[0042] Finally, the initial substation planning is performed according to the empty area set and the shortest path set using a combination algorithm of a greedy algorithm and a dynamic programming.
[0043] Optionally, if the number of available empty area sets is greater than the process system set data, the greedy algorithm is preferentially used for initial substation planning. In an embodiment of the present invention, the empty areas are allocated from large to small, and the process systems with high load calculations are preferentially allocated substations; each process system is individually allocated a substation that meets the proximity principle in each empty space, and the allocation starts from the nearest empty area close to the process system. Each allocation is performed according to the principle of the closest distance to occupy the empty area of the substation. The greedy algorithm adopts the following strategy: the purpose is to allocate all substations in the empty area, and the empty areas are allocated from large to small, and the process systems with high load calculations are preferentially allocated. Each process system is individually allocated a substation that meets the proximity principle in each empty space, and the allocation starts from the nearest empty area close to the system. Each allocation is performed according to the principle of the closest distance, i.e., min{Lfiej}, to occupy the empty area of the substation, i.e., the minimum cable consumption.
[0044] If the number of available empty area sets is less than the process system set data, the dynamic programming strategy is used for the initial substation empty area planning. Optionally, let dp[i][j][k] be the cable consumption of placing the substation corresponding to the i-th process system into the j-th empty area without exceeding the empty area capacity k; determine the state transition process of substation empty area planning, including: if the substation of the i-th process system is not placed into the j-th empty area, then dp[i][j][k] = dp[i-1][j][k]; if the i-th process system is placed into the j-th empty area, then dp[i][j][k] = dp[i-1][j][kw[i]] + v [i]), w[i] is the power supply load of the substation corresponding to the i-th process system, and v[i] is the cable consumption of the substation in the empty area; according to the state transfer process, the state transfer equation is obtained: dp[i][j][k]=min(dp[i-1][j][k],dp[i-1][j][kw[i]]+v[i]); according to the state transfer equation, the initial substation empty area planning corresponding to the minimum cable consumption is solved when the substations corresponding to all process systems are placed in all empty areas.
[0045] If the number of available empty area sets is less than the process system set data, there must be a merger of substations. With the minimum cable consumption as the goal, a dynamic programming matrix is established. The whole problem can be described as: there are {S1, S2, S3....Sn}n process systems, the power load of the i-th process system is Pi, the cable consumption in the j-th empty area is *Lfiej, c is the cable consumption constant term, the area E of the empty area can carry the maximum power supply substation load e*E, e is the empirical constant term obtained based on the power supply load and area, and the solution goal is to obtain the substation planning allocation with the minimum cable consumption without exceeding the maximum power supply load carried by the empty area. The algorithm for solving this problem step by step is as follows:
[0046] Let dp[i][j][k] be the cable consumption required to place the substation corresponding to the i-th process system into the j-th empty area without exceeding the empty area capacity k (the maximum load that the empty area can carry). According to the state transition process:
[0047] If the substation of the ith process system is not placed in the jth empty area, then dp[i][j][k] = dp[i-1][j][k]; if the ith process system is placed in the jth empty area, then dp[i][j][k] = dp[i-1][j][kw[i]] + v[i]), where k>w[i], w[i] is the power supply load of the substation corresponding to the ith process system, and v[i] is the cable consumption of the substation in the empty area. Combining these two situations, the state transfer equation is obtained.
[0048] The boundary conditions of the dynamic programming strategy are: the original state, that is, the case where there is no empty space to carry the substation corresponding to any process system is dp[0][j][k] = 0, that is, for any j and k, the cable consumption is 0. The final goal is min(dp[n][m][C_m]), that is, the minimum cable consumption when all substations corresponding to all process systems are placed in all backpacks. The pseudo code is as follows:
[0049] Pseudocode implementation
[0050]
[0051] At this point, the initial substation planning of the steel plant substation design scheme is completed, and the initial optimized solution of the steel plant substation design scheme is obtained.
[0052] The initial strategy can be used to obtain a substation planning that occupies the most empty space and consumes the least cable, but the initial strategy cannot quantify and calculate other characteristic values of each process system, such as the relationship between project attributes, professions, substations, etc. In addition, when the plant layout changes, it is necessary to re-determine the size, quantity and location of the substation based on the latest empty space situation, and the latest substation design needs to be completed in a short period of time. Therefore, the present invention further introduces a reinforcement learning model. Reinforcement Learning (RL), also known as reinforcement learning, evaluation learning or enhanced learning, is one of the paradigms and methodologies of machine learning, which is used to describe and solve the problem of achieving maximum rewards or achieving specific goals through learning strategies by agents in the process of interaction with the environment. Compared with the common supervised learning and unsupervised learning in machine learning: in the training set of supervised learning, each sample contains a label referring to the correct result, and its task is to let the system infer the appropriate feedback mechanism according to the label corresponding to each sample in the training set, and then calculate a result that is as correct as possible on the sample with unknown label; unsupervised learning is to discover hidden structures from unlabeled data sets, such as cluster analysis. Compared with supervised learning, reinforcement learning does not require a correct "label". Compared with unsupervised learning, which seeks hidden data set structures, the goal of reinforcement learning is to maximize rewards and focus on the interaction between the agent and the environment.
[0053] In step S13, optionally, based on the design data of historical engineering projects, the reinforcement learning-based intelligent agent model is trained with the minimum cable consumption as the goal to obtain a trained reinforcement learning-based intelligent agent model; with the minimum distance between the overall substation and the process system as the reward goal, the initial substation planning is used as the initial state input to the trained reinforcement learning-based intelligent agent model; based on the updated data, the data container is used to synchronize data in real time, the load calculation of the corresponding process system is adjusted, the substation demand is updated by dynamic planning, and the trained reinforcement learning-based intelligent agent model is used to iteratively adjust the design plan until the current project requirements are met to obtain the final design plan.
[0054] See also Figure 2 , the intelligent agent model construction scheme based on reinforcement learning is as follows:
[0055] The core of reinforcement learning is to model decision making and learn the best decision-making plan through interaction with the environment. The framework of reinforcement learning is mainly composed of the following core components: state, action, reward, and strategy. In the present invention, the core components of the intelligent agent model based on reinforcement learning are:
[0056] Status: empty area, process system, pipe network, substation location, specialty, project type and other characteristic value indicators recorded in the current conversion grid;
[0057] Action: The agent places the substation in a certain empty area and connects it to the corresponding process system;
[0058] Bonus: The distance between the substation and its corresponding power supply system, multiplied by a fixed constant term, is the cable consumption. The smaller the value, the better.
[0059] Strategy: It is a mapping function from state to action, usually expressed as pi(a|s) to represent the probability of taking action a in state s. In dealing with the complex application scenarios mentioned in this invention, a multi-layer neural network is used as its mapping function.
[0060] The training method of the reinforcement learning-based agent model is as follows:
[0061] By converting the initial data design drawings and final design drawings of historical engineering projects (i.e., process systems, pipeline networks, and substations) into grid data, the power supply load required by each process system is calculated through the data container, and the first state is obtained by initializing the planning algorithm as the input of the intelligent model based on reinforcement learning. The reward target is to minimize the distance between the overall substation and the process system. If a state in each iterative update meets the state of the final delivered design draft, the iteration is terminated immediately, and a higher reward value is set for the previous strategy function, so as to dynamically optimize its strategy mapping function model.
[0062] The initial state is obtained through the first combinatorial optimization algorithm, and then used as input to the reinforcement learning-based intelligent model to continuously infer a better layout plan. Its overall advantage is that the initial state obtains a numerical optimization solution through the greedy algorithm and the dynamic programming algorithm, and then a solution closer to the actual environment is obtained through the reinforcement learning-based intelligent model obtained based on the design data of historical projects and the reinforcement learning strategy. The reinforcement learning model gives the reinforcement learning-based intelligent model the implicit eigenvalue association relationship in the actual project based on the reward goal of minimum cable consumption and the final design draft.
[0063] When upstream professionals make local data adjustments, such as individual workshop modifications, small equipment adjustments, etc., the embodiment of the present invention will use the data container to synchronize data in real time, adjust the load calculation of the corresponding process system, and dynamically plan and update the substation requirements based on the calculation results, and input the results into the intelligent agent model based on reinforcement learning for iteration, so as to continuously iterate and adjust the design plan until the latest design plan meets the existing requirements of the project. In this way, the final design plan is obtained.
[0064] The core of the embodiment of the present invention is to realize the optimization of the scheme by using the intelligent agent based on data container, combinatorial optimization and reinforcement learning to realize the optimization of the scheme in view of the complex and changeable process flow of the steel plant project and the limited open space in the plant area, so that the adjustment of the substation design scheme can basically achieve intelligentization, and can adapt to various upstream professional data and design changes to quickly iterate and change the scheme. The power load of all equipment in each workshop in the process system is calculated in real time through the data container, and then the data container is connected with the dynamic planning scheme to output the initialization scheme; specifically, the CAD design drawing is converted into a regional placeholder map in grid representation, and the actual distance from each substation to the process system is calculated through the grid placeholder map; by combining the regional placeholder map and the combinatorial optimization algorithm, the greedy algorithm and the dynamic programming algorithm, the initial optimization solution that can be obtained in theoretical numerical terms is obtained. The embodiment of the present invention also combines the calculated initial solution of the historical engineering project with the final solution, and trains the intelligent agent model that can obtain the optimal solution with the goal of minimizing cable consumption through an intelligent agent model based on reinforcement learning; the intelligent agent model not only aims to minimize cable consumption, but also accepts other quantifiable project parameters as feature value inputs, and takes the final design draft as the optimization goal, dynamically adjusts and optimizes its strategy mapping function, and associates information such as project parameters with the final design results, rather than simply aiming to minimize cable consumption. For the initially generated substation results, they are input into the intelligent agent model based on reinforcement learning. The intelligent agent model begins to continue iterating and optimizing according to the current state of the substation data. The intelligent agent model trained with past historical data can continue to complete the optimization according to the current state of the intelligent agent model until the best guess is met and the reasoning is ended.
[0065] In summary, the embodiment of the present invention loads all data required for substation design into a data container, and uses the data container to calculate the total load of each substation; converts the CAD design drawing into an area placeholder map represented by grid blocks by layer, and performs initial substation planning based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm; takes the minimum distance between the overall substation and the process system as the reward goal, uses the initial substation planning as the initial state input to the intelligent model based on reinforcement learning, dynamically adjusts and optimizes the layout plan based on updated data, and obtains the final design plan, thereby solving the problems of excessive design pressure, easy data errors and omissions, and low plan rationality in the substation plan design process, and realizes the intelligent design of the substation plan by using AI technology, effectively ensuring the accuracy, real-time and efficiency of the design plan.
[0066] like Figure 3 The figure shows a structural diagram of a real-time optimization device for a steel plant substation design scheme based on AI technology provided by an embodiment of the present invention, the device comprising:
[0067] The data acquisition module 301 is used to load all data required for substation design into a data container, and use the data container to calculate the total load of each substation;
[0068] The initial planning module 302 is used to convert the CAD design drawing into a regional placeholder map represented by grid blocks according to the layer, and perform initial substation planning based on the regional placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm;
[0069] The dynamic adjustment module 303 is used to take the minimum distance between the overall substation and the process system as the reward target, use the initial substation planning as the initial state input to the intelligent model based on reinforcement learning, and dynamically adjust the optimized layout plan based on the updated planning parameters to obtain the final design plan.
[0070] The device of the above embodiment is applied to the corresponding method in the above embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.
[0071] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor 401, a communication interface 402, a memory 403 and a communication bus 404, wherein the processor, the communication interface and the memory communicate with each other through the communication bus. The processor may call the logic instructions in the memory to execute a real-time optimization method for a steel plant substation design scheme based on AI technology, the method comprising: loading all data required for substation design into a data container, and applying the data container to calculate the total load of each substation; converting the CAD design drawing into a regional placeholder map represented by a grid block by layer, and performing the initial substation planning based on the regional placeholder map using a combination algorithm of a greedy algorithm and a dynamic programming; taking the minimum distance between the overall substation and the process system as the reward target, taking the initial substation planning as the initial state input to the intelligent agent model based on reinforcement learning, dynamically adjusting the optimization layout plan based on the updated data, and obtaining the final design plan.
[0072] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0073] On the other hand, an embodiment of the present invention further provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute a real-time optimization method for a steel plant substation design scheme based on AI technology provided by the above-mentioned method embodiments, the method including: loading all data required for substation design into a data container, and using the data container to calculate the total load of each substation; converting the CAD design drawing into an area placeholder map represented by grid blocks by layer, and performing initial substation planning based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm; taking the minimum distance between the overall substation and the process system as the reward target, using the initial substation planning as the initial state input to an intelligent model based on reinforcement learning, and dynamically adjusting the optimized layout plan based on the updated data to obtain the final design plan.
[0074] On the other hand, an embodiment of the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute a real-time optimization method for a steel plant substation design scheme based on AI technology provided in the above-mentioned embodiments, the method comprising: loading all data required for substation design into a data container, and using the data container to calculate the total load of each substation; converting the CAD design drawing into an area placeholder map represented by grid blocks by layer, and performing initial substation planning based on the area placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm; taking the minimum distance between the overall substation and the process system as the reward target, using the initial substation planning as the initial state input into an intelligent agent model based on reinforcement learning, dynamically adjusting the optimized layout plan based on the updated data, and obtaining the final design plan.
[0075] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.
[0076] The above is only a partial implementation of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A real-time optimization method for steel plant substation design based on AI technology, characterized in that: The method comprises: Loading all data required for substation design into a data container, and using the data container to calculate the total load of each substation; Convert the CAD design drawing into a regional placeholder map represented by grid blocks according to the layer, and perform the initial substation planning based on the regional placeholder map using a combination of a greedy algorithm and a dynamic programming algorithm; Taking the minimum distance between the overall substation and the process system as the reward target, the initial substation planning is used as the initial state input to the reinforcement learning-based intelligent model, and the layout plan is dynamically adjusted and optimized based on the updated data to obtain the final design plan.
2. The method according to claim 1, characterized in that The applying the data container to calculate the total load of the substation includes: The total power load of each process system is calculated by applying the following relationship through the data container, and the total load of each substation is determined by summing up: Among them, Sjs is the total apparent power of the substation, Pjs is the active power of each workshop, Qjs is the reactive power of each workshop, and m is the number of workshops powered by the substation.
3. The method according to claim 1, characterized in that The method of converting the CAD design drawing into a regional placeholder map represented by grid blocks according to the layer, and performing the initial substation planning using a combination algorithm of a greedy algorithm and a dynamic programming according to the regional placeholder map, includes: The CAD design drawing is represented as a process system layer and a pipe network bridge layer according to the layer, and the occupied location information and unoccupied area information of each process system in the current plant area are obtained by converting the pipe network bridge layer and the process system layer into a grid block representation and storing them in a matrix; The unoccupied area information is divided into a usable empty area set and a pipe network occupied area according to the minimum area constraint of the boundary geometry; A path-finding algorithm is applied to form a set of shortest paths from each process system through the pipe network occupied area to each empty area; The initial substation planning is performed according to the empty area set and the shortest path set using a combination algorithm of a greedy algorithm and a dynamic programming.
4. The method according to claim 3, characterized in that The initial substation planning is performed using a combination algorithm of a greedy algorithm and a dynamic programming according to the empty area set and the shortest path set, comprising: If the number of available empty area sets is greater than the process system set data, the greedy algorithm is used first for the initial substation planning; If the number of available empty area sets is less than the process system set data, the dynamic programming strategy is used for the initial substation empty area planning.
5. The method according to claim 4, characterized in that The initial substation planning using the greedy algorithm includes: Start allocating the empty areas from large to small, and give priority to substation allocation for process systems with high load calculations; Each process system is allocated a substation that meets the proximity principle in each open space, starting from the nearest empty area close to the process system. Each allocation is carried out according to the principle of the closest distance to occupy the empty area of the substation.
6. The method according to claim 4, characterized in that The use of dynamic programming strategy to perform initial substation empty area planning includes: Let dp[i][j][k] be the cable consumption required to place the substation corresponding to the i-th process system into the j-th open space without exceeding the capacity of the open space k; The state transition process of determining the substation empty area planning includes: if the substation of the i-th process system is not placed in the j-th empty area, then dp[i][j][k]=dp[i-1][j][k]; if the i-th process system is placed in the j-th empty area, then dp[i][j][k]=dp[i-1][j][kw[i]]+v[i]), w[i] is the substation power supply load corresponding to the i-th process system, and v[i] is the cable consumption of the substation in the empty area; According to the state transfer process, a state transfer equation is obtained: dp[i][j][k]=min(dp[i-1][j][k],dp[i-1][j][kw[i]]+v[i]); The state transfer equation is used to solve the initial substation empty area planning corresponding to the minimum cable consumption when the substations corresponding to all process systems are placed in all empty areas.
7. The method according to claim 1, characterized in that The reward target is to minimize the distance between the overall substation and the process system, and the initial state of the initial substation planning is used as the input of the intelligent agent model based on reinforcement learning. The layout plan is dynamically adjusted and optimized based on the updated data to obtain the final design plan, including: Based on the design data of historical engineering projects, the reinforcement learning-based agent model is trained with the goal of minimizing cable consumption to obtain a trained reinforcement learning-based agent model; An agent model based on reinforcement learning trained with the initial substation planning as the initial state input, with the minimum distance between the overall substation and the process system as the reward target; Based on the updated data, the data container is used to synchronize data in real time, adjust the load calculation of the corresponding process system, dynamically plan and update the substation demand, and apply the trained reinforcement learning-based intelligent model to iteratively adjust the design plan until the current project requirements are met and the final design plan is obtained.
8. A real-time optimization device for steel plant substation design based on AI technology, characterized in that: The device comprises: A data acquisition module, used to load all data required for substation design into a data container, and use the data container to calculate the total load of each substation; The initial planning module is used to convert the CAD design drawing into a regional placeholder map represented by grid blocks according to the layer, and perform the initial substation planning based on the regional placeholder map using a combination of greedy algorithm and dynamic programming algorithm; The dynamic adjustment module is used to take the minimum distance between the overall substation and the process system as the reward target, use the initial substation planning as the initial state input to the reinforcement learning-based intelligent model, and dynamically adjust the optimized layout plan based on the updated planning parameters to obtain the final design plan.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements a real-time optimization method for a steel plant substation design scheme based on AI technology as described in any one of claims 1-7.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, a real-time optimization method for a steel plant substation design scheme based on AI technology is implemented as described in any one of claims 1 to 7.