Method and system for realizing multi-energy flow calculation optimization based on industrial-grade configuration platform

Abstract

This invention discloses a method and system for optimizing multi-energy flow computation based on an industrial-grade configuration platform. The method includes: establishing a power grid and heating network model based on the transmission patterns of different energy media in a multi-energy system, and forming a unified set of computational equations through power balance constraints of coupled nodes. Model parameters and state variables are mapped to tensor data, constructing a directed acyclic tensor flow computation graph composed of multiple nodes, and configuring linear or nonlinear solver interfaces and automatic differentiation techniques for optimization. The computation graph is activated using a data source or event-driven mechanism to achieve parallel scheduling and execution of computational nodes, and the solution results are synchronized to a visualization interface, database, and control unit. Implementing this method effectively reduces model coupling and improves nonlinear solution efficiency, significantly enhancing the computational efficiency of multi-energy system state analysis and greatly increasing configuration flexibility, thus providing possibilities for more intelligent and efficient multi-energy system management.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system technology, and in particular to a method and system for multi-energy flow computation optimization based on an industrial-grade configuration platform. Background Technology

[0002] With the continuous advancement of new power system construction, integrated energy systems, also known as multi-energy systems, have gradually become a research and application hotspot. These systems integrate multiple energy sources such as electricity, heat, cooling, and gas, and their penetration rate on the load side continues to increase. Multi-energy flow computation, as a key technology for evaluating the operating status of such complex systems, performing safety checks, and implementing optimized scheduling, is crucial to the overall system efficiency due to its accuracy and real-time performance. However, because it involves interdisciplinary physical processes such as electromagnetic transients, hydraulic flow, and heat conduction, the mathematical model of multi-energy systems is essentially a vast and complex system of high-dimensional nonlinear algebraic equations.

[0003] Traditional multi-energy flow calculation methods typically rely on dedicated power flow calculation programs custom-developed for specific physical topologies. When faced with changes in system topology or the addition of new energy media, these methods require extensive modifications to the underlying code, resulting in extremely high maintenance costs. Furthermore, due to the highly nonlinear nature of multi-energy flow problems, using the conventional Newton-Raphson method not only consumes significant computational resources for differentiation and Jacobian matrix updates but also struggles to meet the rapid response requirements of real-time energy management systems.

[0004] Therefore, it is necessary to design a new method to effectively reduce model coupling and improve nonlinear solution efficiency. This will not only significantly enhance the computational efficiency of multi-energy system state analysis, but also greatly increase the flexibility of configuration, making it possible to achieve more intelligent and efficient multi-energy system management. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for optimizing multi-energy flow computing based on an industrial-grade configuration platform.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform, comprising: Based on the transmission laws of different energy media in multi-energy systems, calculation models of power grid and heating network are established respectively, and power balance constraints of coupling nodes are established to form a unified and computable set of equations and constraints, so as to obtain a multi-energy flow coupling physical model. All parameters and state variables in the multi-energy flow coupled physical model are mapped to tensor data with a unified structure, and a directed acyclic tensor flow computation graph composed of multiple nodes is configured. Configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and generate standardized descriptions and required matrices for the equation set and tensor data through automatic differentiation techniques to obtain an optimized directed acyclic tensor flow computation graph. The optimized directed acyclic tensor flow computation graph is run using a data source or event-driven mechanism. The computation nodes are sorted and scheduled to achieve parallel execution. The solution results are then mapped to a visualization component, database, or control unit.

[0007] This invention also provides a multi-energy flow computing optimization implementation system based on an industrial-grade configuration platform, comprising: The building unit is used to establish calculation models of power grid and heating network based on the transmission laws of different energy media in multi-energy systems, and to establish power balance constraints of coupling nodes, forming a unified and computable set of equations and constraints to obtain a multi-energy flow coupling physical model. The conversion unit is used to map all parameters and state variables in the multi-energy flow coupled physical model into tensor data with a unified structure, and to configure a directed acyclic tensor flow computation graph composed of multiple nodes. An optimization unit is used to configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and to generate a standardized description and required matrix for the equation set and the tensor data through automatic differentiation technology to obtain an optimized directed acyclic tensor flow computation graph. The solution mapping unit is used to run the optimized directed acyclic tensor flow computation graph using a data source or event-driven mechanism, sort and schedule the computation nodes to achieve parallel execution, and then map the solution results to a visualization component, database or control unit.

[0008] The advantages of this invention compared to existing technologies are as follows: By implementing an optimized multi-energy flow calculation method on an industrial-grade configuration platform, this invention effectively reduces model coupling and significantly improves nonlinear solution efficiency. Specifically, firstly, a unified power grid and heating network calculation model is established based on the transmission laws of different energy media in a multi-energy system, and power balance constraints are established to form a set of equations. Then, these model parameters are mapped into tensor data structures, constructing a directed acyclic tensor flow calculation graph composed of multiple nodes, thus completely decoupling the calculation engine from problem modeling. Next, solver interfaces are configured in the solution nodes of the calculation graph, and standardized descriptions and required matrices are generated through automatic differentiation technology, thereby accelerating the solution process. Finally, the optimized calculation graph is run using a data source or event-driven mechanism to achieve efficient sorting, scheduling, and parallel execution of calculation nodes. The results can be easily mapped to visualization components, databases, or control units, which not only significantly enhances the calculation efficiency of multi-energy system state analysis but also greatly increases configuration flexibility, providing a solid foundation for achieving more intelligent and efficient multi-energy system management.

[0009] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart illustrating the multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform provided in this embodiment of the invention. Figure 1 ; Figure 2 A flowchart illustrating the multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform provided in this embodiment of the invention. Figure 2 ; Figure 3 This is a schematic diagram of the structure of tensor data provided in an embodiment of the present invention; Figure 4 A topology diagram of an integrated electric-thermal energy system provided in an embodiment of the present invention; Figure 5 A diagram showing the multi-energy flow calculation results of the topology of the integrated electric-thermal energy system provided in an embodiment of the present invention; Figure 6 This invention provides a multi-energy flow tensor flow computation graph architecture built on an industrial-grade configuration platform. Figure 7This is a multi-energy optimized configuration diagram built on an industrial-grade configuration platform, provided in an embodiment of the present invention. Figure 8 This is a visualization analysis chart of the multi-energy flow calculation and multi-energy optimization scheduling results provided in the embodiments of the present invention; Figure 9 This is a schematic block diagram of a multi-energy flow computing optimization implementation system based on an industrial-grade configuration platform, provided in an embodiment of the present invention. Figure 10 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0014] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0015] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0016] Please see Figure 1 and Figure 2 , Figure 1 A flowchart illustrating the multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform provided in this embodiment of the invention. Figure 1 . Figure 2 A flowchart illustrating the multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform provided in this embodiment of the invention. Figure 2This multi-energy flow computation optimization method, based on an industrial-grade configuration platform, is applied to a server. By establishing power grid and heating network models and defining power balance constraints at coupled nodes, a unified set of equations and constraints is formed to construct a coupled physical model of multi-energy flows. Furthermore, all parameters and state variables are mapped to tensor data, and a directed acyclic tensor flow computation graph is configured. Automatic differentiation techniques are used to generate standardized descriptions and required matrices, optimizing the solution process. The computation graph is run using a data source or event-driven mechanism, enabling the sorting, scheduling, and parallel execution of computation nodes. Finally, the solution results are mapped to visualization components, databases, or control units. This method effectively reduces the coupling between models, improves the efficiency of nonlinear solutions, enhances the computational efficiency of multi-energy system state analysis, and significantly improves configuration flexibility, providing possibilities for intelligent and efficient multi-energy system management.

[0017] Figure 2 This is a flowchart illustrating the multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform provided in an embodiment of the present invention. Figure 2 As shown, the method includes the following steps S110 to S140.

[0018] S110. Based on the transmission laws of different energy media in a multi-energy system, calculation models for the power grid and heating network are established respectively, and power balance constraints of the coupling nodes are established to form a unified and computable set of equations and constraints, so as to obtain a multi-energy flow coupling physical model.

[0019] In this embodiment, the multi-energy flow coupling physical model refers to a comprehensive mathematical model that reflects the dynamic behavior of the entire multi-energy system by integrating the physical characteristics of the power grid and the heating network and the energy conversion relationships between them. This model includes not only the independent mathematical descriptions of each subsystem, but also the energy exchange rules between them implemented through specific equipment (such as combined heat and power units).

[0020] A set of equations refers to a collection of mathematical expressions comprised of the aforementioned power grid calculation model, heating network calculation model, and the power balance constraints of the coupling nodes between the two. These equations aim to accurately describe the transmission patterns of electricity and heat within their respective networks and their interaction mechanisms.

[0021] A constraint set refers to a set of restrictions established to ensure the physical rationality of energy transfer processes within and between subsystems. For example, Kirchhoff's laws apply in electrical networks, and the principles of conservation of mass and energy are observed in heating networks.

[0022] In one embodiment, step S110 described above may include steps S111 to S114.

[0023] S111. Based on the physical laws of power transmission and Kirchhoff's laws, construct a power grid calculation model that includes nodal voltage equations for the electrical network within a multi-energy system.

[0024] In this embodiment, the power grid calculation model refers to the modeling of a DC microgrid based on Kirchhoff's Current Law (KCL) and Voltage Law (KVL), combined with the branch volt-ampere characteristics. Its core is the node voltage equation, which is used to describe the relationship between voltage and current at each point in the network.

[0025] S112. Based on the principle of mass and energy conservation, a hydraulic and thermal equation model is developed for the heating network to obtain the calculation model of the heating network.

[0026] In this embodiment, the heating network calculation model refers to the model constructed using the principle of mass and energy conservation of the heating network, which includes the hydraulic head loss equation and the thermal temperature distribution equation, to simulate the flow of the heat medium and the heat transfer process in the pipeline.

[0027] S113. In a multi-energy system, different energy networks are connected through energy conversion equipment. Based on the input and output characteristics of the energy conversion equipment, power balance constraints of the coupling nodes are established.

[0028] In this embodiment, the power balance constraint of the coupling node refers to the requirement that for key nodes such as combined heat and power plants, the electrical energy consumption and thermal energy production must be precisely matched, that is, the power consumption at the node plus the generated thermal energy equals the total energy input through the energy conversion equipment.

[0029] S114. The power grid calculation model and the heating network calculation model, together with the power balance constraints of the coupled nodes, are organized into a comprehensive set of equations and constraints to obtain a multi-energy flow coupled physical model.

[0030] This step emphasizes integrating the mathematical models of the power grid and heating network with their coupling relationships to form a complete and consistent multi-energy flow coupled physical model, which facilitates subsequent analysis and optimization.

[0031] In this embodiment, based on the transmission physical laws of different energy media within the multi-energy system, a power grid calculation model containing node voltage equations and a heating network calculation model containing hydraulic and thermal equations are established respectively; power balance constraints of coupled nodes are established based on the input-output relationship of energy conversion equipment; the node sets, branch sets, equipment parameter sets and boundary conditions of each sub-network are uniformly organized into a computable set of equations and constraints, which serve as the input for subsequent tensorization and solution.

[0032] Specifically, in constructing the multi-energy flow coupling physical model, for a system where a DC microgrid is coupled to a heating network, the power coupling constraint is associated through the nodes of the cogeneration unit, satisfying the physical conservation relationship that the heat generation power of the node is equal to the sum of the power consumption power of the node and the output power of the cogeneration unit. In this process, by reading the grid node parameters, heating network parameters, and the rated capacity and operating boundary of the cogeneration unit, a grid sub-model, a heating network sub-model, and a coupling equipment sub-model are formed respectively, and then the three are combined by the power balance constraint of the coupling nodes.

[0033] The model of the DC microgrid is constructed based on Kirchhoff's current law, Kirchhoff's voltage law, and branch volt-ampere characteristics, as shown in the following equation: In the formula: A e For the microgrid node branch correlation matrix; I b I is the branch current vector; s V is the current vector of the branch current source; b V is the branch voltage drop vector; n V is the node voltage vector; s This is the voltage vector of the branch voltage source.

[0034] The voltage at the connection point between the DC microgrid and the grid is relatively stable. The node-to-ground branch can be represented by a voltage source, and the DC load can be represented by a current source on the node-to-ground branch. Eliminating the branch current and voltage difference variables in the model, we obtain the node voltage equation as shown below: ; The heating network model includes a hydraulic model and a thermal model, as shown in the following equation: In the formula: A h The branch correlation matrix of the heating network nodes; m b h is the branch heat medium flow rate vector; b h is the branch head loss vector; n L is the nodal pressure vector; b and D b These are the branch pipe length and diameter vectors, respectively; μ is the flow viscosity; ρ is the heat transfer medium density; g is the gravitational acceleration; C is the specific heat capacity of the heat transfer medium; T b Let be a diagonal matrix with the temperature difference between the two ends of each branch as its diagonal element; λ is the thermal conductivity of the pipe; m i,k and m o,k These represent the heat medium flow rates at the inflow and outflow nodes, respectively; T i,k and T o,k These are the temperatures of the heat medium at the inflow and outflow nodes, respectively.

[0035] At the node where the fuel cell cogeneration unit is located, the heat generation power must equal the sum of the node's power consumption and the fuel cell's output power. In the formula: m fT represents the heat transfer fluid flow rate at the fuel cell node. fo and T fi These are the temperatures of the heat transfer medium flowing into and out of the fuel cell node, respectively; V f and I f These represent the voltage of the fuel cell node and the injection current into the DC microgrid, respectively; P f This refers to the output power of the fuel cell.

[0036] The above power grid calculation model and heating network calculation model, along with the power balance constraints of the coupled nodes, are integrated to form a comprehensive set of equations and constraints.

[0037] These equations and constraint sets serve as inputs for subsequent tensor quantization and solving, ensuring that the dynamic behavior of the entire multi-energy system is accurately reflected.

[0038] In one specific embodiment, for a multi-energy system consisting of a DC microgrid and a heating network, the coupling point achieves the conversion of electrical energy into heat energy through a combined heat and power (CHP) unit. The specific operation steps are as follows: Read grid node parameters: including grid node number, injected current, voltage and other information.

[0039] Read heating network parameters: including heating network node number, heat medium flow rate, temperature and other data.

[0040] Determine the rated capacity and operating boundaries of the combined heat and power (CHP) unit: Understand the unit's maximum output power and its operating limitations.

[0041] Forming sub-models: Based on the above information, form sub-models for power grid, heating network, and coupling equipment respectively.

[0042] Simultaneous Model: The three sub-models are combined by coupling node power balance constraints to form a complete set of multi-energy flow nonlinear equations, which serve as the basis for subsequent graphical configuration and solution.

[0043] Specifically, for various energy networks (such as distribution networks and heating networks) and their coupling hubs (such as cogeneration units) within a multi-energy system, voltage equations for electrical network nodes based on Kirchhoff's laws and hydraulic and thermal models for the heating network based on mass and energy conservation are constructed respectively. Furthermore, the rated parameters, boundary conditions, and initial values ​​of each node, branch, and conversion device are read, and the mathematical models of different energy networks are combined through the power balance constraints of the coupling nodes to form a set of multi-energy flow nonlinear equations, providing a unified model input for subsequent graphical configuration and solution.

[0044] This method can not only effectively simulate and analyze the behavior of multi-energy systems, but also provide a scientific basis for optimized scheduling, thereby improving the system's operating efficiency and economy.

[0045] S120. Map all parameters and state variables in the multi-energy flow coupled physical model to tensor data with a unified structure, and configure a directed acyclic tensor flow computation graph composed of multiple nodes.

[0046] In this embodiment, the directed acyclic tensor flow computation graph refers to a data processing architecture based on graphical representation. It describes the flow of data and processing logic by defining different types of nodes (such as data acquisition, matrix operation, and solution nodes) and the connection relationships between them, ensuring that data flows and is processed efficiently according to predetermined dependencies.

[0047] In one embodiment, step S120 described above may include steps S121 to S123.

[0048] S121. Determine all key parameters and state variables that need to be solved in the multi-energy flow coupled physical model, and define the roles of key parameters and state variables in the multi-energy flow model to obtain the definition results.

[0049] In this embodiment, defining the result means identifying and clarifying the roles of all key parameters and state variables in the multi-energy flow model, providing a basis for subsequent mapping to tensor structures, including determining which are inputs, outputs, or intermediate variables.

[0050] Specifically, key parameters are fixed or variable values ​​that describe system characteristics, such as branch resistance, conductivity matrix, and heat transfer fluid flow rate.

[0051] State variables: These are dynamic variables that need to be calculated, such as node voltages and heat transfer medium temperatures. They represent the current state of the system.

[0052] Role definition: Clarify the role of each parameter or state variable in the multi-energy flow model. For example, some parameters may serve as inputs (boundary conditions), while others serve as outputs (solution results).

[0053] S122. Based on dimensional planning, map the parameters and state variables in the defined results to tensor data structures, and reconstruct them into tensor groups by combining them with topological data to obtain tensor data.

[0054] In this embodiment, as Figure 3 As shown, tensor data refers to mapping these parameters and state variables to a unified multidimensional array structure based on their dimensional characteristics, forming a high-dimensional tensor group that supports automatic addressing, slice calling, and parallel operation, so as to facilitate efficient numerical calculation and optimization on an industrial-grade configuration platform.

[0055] Specifically, the dimensional allocation of parameters and state variables in the tensor is determined based on their characteristics, such as time dimension, node dimension, branch dimension, etc.

[0056] Tensor Reconstruction: Utilizing multidimensional arrays of tensors, various topological and state data in a multi-functional system are transformed into high-dimensional tensor sets suitable for parallel computation at the platform's underlying layer. Specifically, this includes, but is not limited to: Branch correlation matrix (A) e and A h ); Branch conductance matrix (Y b ); Heat medium flow rate vector (m) b ); Node pressure vector (h) n ); Configuration properties: Set dimension labels, index mapping relationships, and data type identifiers for each tensor to support automatic addressing, slice invocation, and data transfer between nodes.

[0057] In summary, this method uses tensor forms to uniformly represent system parameters and states, and achieves modular modeling and rapid solution of complex systems through carefully designed computational graphs.

[0058] Specifically, the data type is extended by using the multidimensional array form of tensors. The topological and state data such as branch correlation matrix, branch conductivity matrix, and heat medium flow rate vector in the multi-energy system are combined with the name string that identifies the length of the dimension and reconstructed into tensor groups suitable for parallel matrix operations at the underlying level of the platform. Dimension labels, index mapping relationships and data type identifiers are configured for each tensor to support automatic addressing, slice calling and inter-node transfer within the configuration platform.

[0059] The node types of the directed acyclic tensor flow computation graph include data acquisition nodes, tensor conventional computation nodes, and solution nodes for nonlinear equation systems or mixed integer programming; the edges of the directed acyclic tensor flow computation graph are used to define the transmission path of tensor data between the first and last nodes and the aggregation and transpose preprocessing operations.

[0060] S123. Based on the computational dependencies between the variables of the tensor data, configure a directed acyclic tensor flow computation graph consisting of data acquisition nodes, matrix operation nodes, constraint generation nodes and solution nodes in the industrial-grade configuration platform, and define the tensor reading, splicing, transposition, aggregation and distribution paths through the edges between nodes.

[0061] Node type: Data acquisition node: Responsible for reading initial data from external sources or updating data in real time.

[0062] Tensor regular computation nodes: perform basic mathematical operations, such as addition, multiplication, transpose, etc.

[0063] Nonlinear equation system or mixed integer programming solution node: Specialized to handle complex equation solving tasks.

[0064] The function of edges is to define the tensor data flow path between the first and last nodes, as well as necessary preprocessing operations (such as aggregation and transpose).

[0065] Computational Graph Configuration: Based on the computational dependencies between tensor data, a directed acyclic graph (DAG) consisting of the aforementioned different types of nodes is configured on an industrial-grade configuration platform. This graph not only clarifies how data flows between different nodes but also specifies the concrete computational logic for each step, ensuring that the entire process proceeds efficiently and systematically.

[0066] These three steps effectively transform the multi-energy-flow coupled physical model into an easily manageable and scalable computational framework. This framework fully leverages the advantages of modern computing resources to achieve fast and accurate simulation and optimized scheduling. This approach significantly improves the development efficiency and flexibility of traditional manual programming methods, making the modeling of complex systems more intuitive and convenient.

[0067] In this embodiment, the known device parameters, topological connections, boundary conditions, and state variables to be determined in the multi-energy flow model are mapped into a unified tensor data structure according to the node dimension, branch dimension, device dimension, and time dimension. Based on the computational dependencies between variables, a directed acyclic tensor flow computation graph consisting of data acquisition nodes, matrix operation nodes, constraint generation nodes, and solution nodes is configured in the industrial-grade configuration platform, and the reading, splicing, transposing, aggregation, and distribution paths of tensors are defined through the edges between nodes. This involves using tensor data structures to uniformly encapsulate node parameters, branch attributes, and topological correlation matrices in a multi-energy system. On an industrial-grade configuration platform, a directed acyclic tensor flow computation graph is constructed using a graphical configuration method. This graph consists of nodes responsible for different mathematical operations (such as matrix transformations and nonlinear solutions) and edges responsible for data flow, thus transforming complex equation solving into a clear, modular data flow. In the state representation process based on tensor structures, the data type is extended by using the multidimensional array form of tensors. The topological and state data such as branch association matrix, branch conductivity matrix, and heat medium flow rate vector in the system are reconstructed into a high-dimensional tensor group suitable for parallel matrix operations at the underlying level of the platform by combining the name string that identifies the dimension length. Dimension labels, index mapping relationships and data type identifiers are configured for each tensor to support automatic addressing, slice calling and inter-node transfer within the configuration platform.

[0068] The device parameters, network node correlation matrices, and undetermined state variables in the multi-energy flow model are transformed into a unified high-order tensor data structure. The data type is expanded using multidimensional arrays of tensors, and the branch correlation matrices in the system are then transformed. and Branch conductance matrix heat medium flow rate vector The topological and state data, combined with name strings indicating the length of the identifier dimension, are reconstructed into a high-dimensional tensor set suitable for parallel matrix operations at the platform's underlying level. Based on this, the multi-energy flow mathematical solution process is mapped into a directed acyclic tensor flow computation graph composed of nodes and edges within the industrial-grade configuration platform. The node types of this computation graph include data acquisition nodes, conventional tensor computation nodes, and nonlinear equation solving nodes. The edges between nodes define the transmission paths of tensor data and data preprocessing operations such as aggregation and transposition, thereby transforming complex equation solving into a structured modular data flow.

[0069] Therefore, in multi-energy systems, a tensor-based encapsulation method is employed for efficient computation and optimization. This method encapsulates key information such as system node parameters, branch attributes, and topological correlation matrices into a high-dimensional tensor set. This encapsulation not only enables the unified processing of different types of physical quantities (such as electrical and thermal quantities) but also fully leverages the advantages of tensor operations, supporting automatic addressing, slice calling, and parallel computing.

[0070] Specifically, on an industrial-grade configuration platform, a directed acyclic tensor flow computation graph is constructed graphically. This computation graph consists of nodes responsible for specific mathematical operations (such as matrix transformations and nonlinear solutions) and edges defining the direction of data flow. Each node represents a specific computational task or data transformation process, while the edges represent the data flow paths between these nodes. This design allows us to transform complex equation-solving problems into a series of clear, modular data flow operations, thereby simplifying the entire solution process and improving computational efficiency.

[0071] During state representation, the multidimensional array properties of tensors are utilized to expand data types. This means that important topological and state data, such as branch correlation matrices, branch conductance matrices, and heat medium flow rate vectors, are reconstructed into high-dimensional tensors suitable for parallel computation at the platform's underlying layer. Simultaneously, each tensor is configured with detailed dimension labels, index mapping relationships, and data type identifiers. This not only improves the accuracy of data processing but also facilitates efficient data access and manipulation within the configuration platform.

[0072] Ultimately, all device parameters, network node association matrices, and state variables to be solved in the multi-energy flow model are transformed into a unified high-order tensor data structure. Based on this, a carefully designed directed acyclic tensor flow computation graph can transform the complex mathematical solution process of multi-energy flow into a structured modular data flow. This not only makes the solution process more intuitive and easier to understand but also greatly improves computational efficiency and flexibility, laying a solid foundation for subsequent analysis and optimization.

[0073] In this embodiment, the computational process involved in the cyberphysics of multi-energy systems has the following characteristics: Different states may be interconnected; for example, the energy flow state requires the use of the topological state, and the photovoltaic power generation prediction state requires the use of the weather prediction state. The algorithms for state calculation may involve a variety of mathematical problems, such as solving systems of equations (e.g., energy flow calculation), solving optimization models (e.g., optimal scheduling, risk assessment), and neural networks (e.g., load forecasting). Cyber-physical analysis and related operations of multi-energy systems can all be completed using tensor computation.

[0074] Based on the above characteristics, the structure of tensor data is designed as follows: Figure 3 As shown.

[0075] S130. Configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and generate standardized descriptions and required matrices for the equation set and tensor data through automatic differentiation techniques to obtain an optimized directed acyclic tensor flow computation graph.

[0076] In this embodiment, the optimized directed acyclic tensor flow computation graph refers to the process of automatically generating the derivative information required for solving the problem by configuring appropriate linear or nonlinear solver interfaces and utilizing automatic differentiation techniques. This transforms complex equations and constraints into a standard input format that the solver can directly process. This optimization enables the computation graph to not only solve multi-energy flow models efficiently and accurately, but also enhances the system's flexibility and adaptability to cope with different computational tasks and application scenarios.

[0077] In one embodiment, step S130 described above may include steps S131 to S134.

[0078] S131. Configure linear or nonlinear solver interfaces in the solver nodes of the directed acyclic tensor flow computation graph.

[0079] In industrial-grade configuration platforms, linear or nonlinear solver interfaces are configured within the solver nodes of the directed acyclic tensor flow computation graph. These interfaces allow complex mathematical equations and tensor data to be directly passed to appropriate solvers, whether built-in or third-party provided, thereby achieving efficient numerical solutions.

[0080] S132. The system of equations and the tensor data are taken as input and passed to the parsing engine to perform lexical parsing on the formula string, identify variables and match their indices to form an expression tree.

[0081] In this embodiment, the expression includes a standardized description of the objective function, constraint functions, and upper and lower bounds of the variables.

[0082] The system of equations constituting the state characterization of the multi-energy system, along with related tensor data, is input into the analytical engine. The engine first performs lexical analysis on the formula string, identifying all involved variables and matching their corresponding index information based on their positions. Next, based on these analysis results, an expression tree is constructed. This tree structure accurately represents the logical relationship between the objective function, constraints, and boundary conditions.

[0083] S133. Based on the expression tree, the first and second derivatives required in the solution process are automatically generated using automatic differentiation technology.

[0084] Based on the expression tree generated in the previous step, automatic differentiation techniques are used to automatically generate the first derivatives (such as the Jacobian matrix) needed in the solution process, as well as the second derivatives (such as the Hessian matrix) when necessary. This step is crucial for improving the stability and speed of solving nonlinear equations because it avoids the errors introduced by traditional numerical difference methods and can handle the dynamic characteristics of complex systems more efficiently.

[0085] S134. Encapsulate the first and second derivatives, along with the constraint boundary conditions, initial values, and convergence criteria, into a standardized input format required by the solver to obtain the optimized directed acyclic tensor flow computation graph.

[0086] Finally, the first and second derivatives obtained through automatic differentiation, along with parameters such as constraint boundary conditions, initial values, and convergence criteria, are encapsulated into a standard input format that meets the requirements of a specific solver. This approach not only simplifies the data preparation stage in the solution process but also enhances the flexibility and adaptability of the entire computational framework, enabling it to quickly adjust the solution strategy according to different application scenarios.

[0087] In summary, by following these four steps—configuring the solver interface, parsing and constructing the expression tree, applying automatic differentiation techniques, and encapsulating the input into a standardized format—the computational graph of directed acyclic tensor flow can be effectively optimized, thereby significantly improving the efficiency and accuracy of solving multi-energy flow models. This workflow design fully considers the needs of practical engineering applications, achieving seamless integration from physical model definition to numerical solution, greatly facilitating the operation for engineers and technicians.

[0088] In this embodiment, firstly, an appropriate linear or nonlinear solver interface is configured in the computation graph, and the defined tensor data and equation expressions are input into the parsing engine. The parsing engine is responsible for performing lexical analysis and variable index matching on the equation string, constructing an expression tree, and generating a standardized description of the objective function, constraints, and variable upper and lower bounds. Next, through automatic differentiation, the system can automatically generate and update the Jacobian matrix required during the iterative solution process, as well as the Hessian matrix when needed. This step converts the problem model into a standard input format that the solver can directly call, thereby achieving efficient solving.

[0089] Secondly, a mapping relationship between variable names, tensor indices, and solution variable numbers is pre-established within the configuration platform. When the nonlinear hydraulic head loss equation, thermal temperature distribution equation, and coupled node power balance equation in the heating network are input, these equations undergo syntax parsing to generate an expression tree. Based on this expression tree, the system performs automatic differentiation to obtain the first-order derivative information of each constraint with respect to the solution variables, and further obtains the second-order derivative information when using a second-order algorithm. Finally, this derivative information, along with constraint boundaries, initial values, and convergence criteria, is encapsulated into the solver's input parameters. This process significantly reduces the iterative computation time when solving complex multi-energy networks simultaneously, improving the overall solution efficiency. Through the above methods, not only is the system's flexibility and adaptability enhanced, but the accuracy and efficiency of the solution process are also guaranteed.

[0090] Specifically, linear or nonlinear solver interfaces are configured in the computing nodes of the configuration platform, eliminating the need for customizing the solver environment and refactoring code based on specific programming languages. This is particularly relevant when dealing with equations involving nonlinear hydraulic head loss in heating networks. When solving multi-energy network simultaneous problems involving coupled node nonlinear power balance equations, the configuration platform first performs a unified analysis of the device models, energy flow relationships, control logic, and optimization objectives configured by the user in the graphical interface. This transforms nodes, branches, devices, variables, parameters, and constraint rules into independent and callable standardized computational units. Each standardized computational unit is encapsulated using unified data and function interfaces, separating the model definition layer, business logic layer, and underlying numerical computation engine. This allows the upper-layer configuration to be independent of specific solver implementations, and the underlying computational engine to flexibly switch between power flow calculation engines, dynamic simulation engines, or optimization solution engines based on the computational task type. Based on this decoupled structure, the platform can achieve unified scheduling and collaborative invocation between different computational tasks without changing the model representation in the configuration interface, improving model reusability and system scalability.

[0091] Furthermore, the platform generates corresponding mathematical expressions, constraint expressions, and objective function expressions based on the configuration, and converts them into computable expression trees or intermediate representation structures. On this basis, it constructs computational paths according to variable dependencies, automatically identifying the mapping relationships between state variables, control variables, disturbance variables, and parameters to be optimized. For nonlinear functions involving power flow balance, energy conservation, equipment operation constraints, and economic indicators, the platform performs layer-by-layer differentiation along the computational path using an automatic differentiation mechanism, automatically obtaining the first-order derivative information of the objective function and constraint functions with respect to decision variables. When needed, it can further generate corresponding higher-order derivative information or approximate second-order information for the construction of the Jacobian matrix, gradient vector, and Hessian matrix. Compared to manually deriving partial derivatives or using numerical difference approximation, this method reduces the workload of model reconstruction, minimizes truncation and rounding errors caused by improper selection of difference step sizes, and improves the accuracy and consistency of the differentiation process in complex multi-energy flow models.

[0092] During the solution process, the underlying computing engine, based on the automatically generated derivative information, calls the corresponding Newton-Raphson iteration algorithm, sequential quadratic programming algorithm, interior-point method algorithm, or other gradient-type optimization algorithms to complete state estimation, power flow solution, parameter identification, or optimized scheduling calculation. When the equipment parameters, topology, or control strategy in the configuration model are adjusted, the platform only needs to update the corresponding computing unit and its associated expression, and automatically reconstruct the local computing path and derivative chain, without the need for manual rewriting of the overall differentiation process. After the calculation is completed, the state variables, sensitivity information, and optimization results output by the underlying engine are mapped back to the equipment objects, measurement point labels, and control interfaces in the configuration platform to realize an integrated processing flow of "graphical configuration modeling—computation engine call—automatic differential differentiation—result writeback," thereby improving the modeling efficiency, solution efficiency, and engineering adaptability of multi-energy flow computation optimization.

[0093] S140. Utilize a data source or event-driven mechanism to run the optimized directed acyclic tensor flow computation graph, sort and schedule the computation nodes to achieve parallel execution, and then map the solution results to a visualization component, database, or control unit.

[0094] In one embodiment, step S140 described above may include steps S141 to S143.

[0095] S141. Set up multiple triggering mechanisms to activate the corresponding computation graph branch in the optimized directed acyclic tensor flow computation graph when real-time data updates or system parameter changes are detected.

[0096] This step primarily involves setting up various triggering mechanisms to activate corresponding branches in the computation graph when real-time data updates or changes in system parameters are detected. Specifically, these triggering mechanisms include: Time-driven: Automatically performs multi-energy flow state calculations and optimized scheduling according to a preset time period.

[0097] Data-driven: When real-time measurement data, boundary conditions, or price parameters change and reach a set threshold, the corresponding calculation process is triggered.

[0098] Event-driven: When specific events such as device startup or shutdown, switch state switching, fault alarm, or control command issuance occur, the relevant calculation process is triggered.

[0099] S142. The activated computation graph branches automatically sort and schedule computation nodes according to the topology hierarchy, and allocate independent threads to subgraphs without data dependencies to achieve parallel processing, so as to perform operations on each computation node. Among them, each computation node performs matrix transformation, nonlinear solution and other operations in the scheduling order to complete the solution of the multi-energy flow model and output energy allocation results and equipment control commands.

[0100] Once the computation graph is activated, the next step is to automatically sort and schedule the activated branches according to their topological hierarchy. During this process, subgraphs without data dependencies are allocated independent threads for parallel processing, thereby improving computational efficiency. Each computation node performs matrix transformations, solves nonlinear equations, and other operations in the scheduled order, ultimately solving the entire multi-energy flow model and outputting energy allocation results and device control commands.

[0101] S143. Synchronously map the energy allocation results, computing node status variables and device control commands to the visualization interface for display, store them in the historical database for analysis, and send them to the underlying control unit for physical device regulation.

[0102] The various triggering mechanisms include time-driven, data-driven, and event-driven mechanisms configured according to operational requirements. Among them, time-driven mechanisms are used to perform multi-energy flow state calculations and optimization scheduling according to a preset cycle; data-driven mechanisms are used to trigger calculations when changes in real-time measurement data, boundary conditions, or price parameters reach a set threshold; and event-driven mechanisms are used to trigger the calculation process when equipment starts or stops, switches between states, fault alarms are triggered, or control commands are issued.

[0103] The final step is to synchronously map the calculated energy allocation results, the state variables of each computing node, and the device control commands to different application layers: Visualized interface display: The energy distribution results and system operating status are displayed intuitively through a graphical interface, which facilitates monitoring and management.

[0104] Historical database storage: The calculation results are saved to a historical database for subsequent analysis, supporting long-term trend analysis and decision-making.

[0105] Low-level control unit regulation: The generated equipment control commands are sent to the low-level control system for direct adjustment and control of physical equipment, ensuring that the actual operation of the system meets the optimization objectives.

[0106] Through the above steps, a closed-loop dynamic response mechanism from data acquisition and processing to result output is realized, which greatly improves the multi-energy system's rapid response and real-time control capabilities in the face of load fluctuations, energy price changes, and changes in operating modes.

[0107] Specifically, by employing data source-driven, event-driven, or time-driven mechanisms, the configuration platform can automatically activate the corresponding directed acyclic tensor flow computation graph branch when the system detects changes in external real-time measurement data, fluctuations in system load, or changes in energy prices. This mechanism ensures that the computation process can be triggered promptly whenever needed.

[0108] Once a computation graph branch is activated, the platform intelligently sorts and schedules the computation nodes according to their internal topology hierarchy. For subgraph parts without data dependencies, the system allocates independent computation threads to process these tasks in parallel, thereby accelerating the solution process of the entire multi-energy flow model. This not only allows for the rapid acquisition of key parameters such as node voltage, heat transfer fluid flow rate, and temperature distribution, but also enables efficient optimized scheduling to respond to dynamic changes in the system.

[0109] Ultimately, all calculation results—including energy allocation schemes, state variables of each node, and device control commands—are synchronously mapped to different application layers: a visual interface displaying calculation results and the current system status for easy monitoring by operators; a historical database storing results for subsequent analysis, supporting long-term trend analysis and decision support; and a low-level control unit directly sending control commands to physical devices to ensure that actual operation aligns with optimization goals. This approach not only improves the responsiveness and flexibility of multi-energy systems in the face of external changes but also enhances the system's automation and intelligence, achieving closed-loop management from data acquisition and real-time calculation to result output. This method effectively supports the efficient operation and management of multi-energy systems.

[0110] After the aforementioned directed acyclic tensor flow computation graph is constructed, the configuration platform configures three types of triggering mechanisms—time-driven, data-driven, and event-driven—according to operational requirements. Time-driven mechanisms are used to perform multi-energy flow state calculations and optimized scheduling according to a preset cycle. Data-driven mechanisms trigger calculations when changes in real-time measurement data, boundary conditions, or price parameters reach a set threshold. Event-driven mechanisms trigger calculation processes when equipment starts / stops, switches between states, fault alarms, or control commands are issued. The platform first receives input data from sensors, controllers, databases, or upper-level scheduling systems, and identifies the affected state variables, constraints, and objective function terms based on a pre-established mapping relationship between data tags and computation nodes. Subsequently, it extracts computation branches related to the changes from the directed acyclic tensor flow computation graph, performs incremental recalculation on the affected subgraphs, and retains the intermediate results from the previous moment for unaffected computation nodes to reduce the computational overhead caused by repeated solutions across the entire graph.

[0111] Furthermore, the platform's underlying layer hierarchically sorts each computing node according to the topological dependencies of the directed acyclic tensor flow computation graph. Nodes without sequential dependencies or data coupling are divided into sets of tasks that can be executed in parallel, and these sets of tasks are scheduled to be executed concurrently in different computing threads, processor cores, or container instances. For key nodes involving nonlinear equation solving, constraint verification, and optimization scheduling, tensor operations, Jacobian updates, constraint discrimination, and solver calls are completed in a preset execution order to obtain results such as node voltage, branch power flow, heat medium flow, temperature distribution, and equipment output.

[0112] After the calculation is completed, the platform writes the output results into the real-time database, historical database and visual configuration interface according to the variable mapping relationship. This is used to complete the operation status display, trend curve refresh, alarm judgment and report generation. At the same time, based on the optimization results, the platform automatically generates equipment control commands or operation setpoints and sends them to converters, pumps, valves, heat exchangers or other controlled terminals through the communication interface. This forms a closed-loop dynamic response mechanism of "data acquisition - status perception - parallel computing - result output - control execution" to improve the calculation response speed and scheduling real-time performance of multi-energy systems under load fluctuations, energy price changes and operation mode switching scenarios.

[0113] Considering energy prices, energy flow balance, and equipment operating constraints, a multi-energy optimization scheduling model is established with the goal of minimizing energy costs. Optimization calculations are triggered when system load or energy prices change. The model is shown in the following equation: In the formula: c g For electricity price; P g Power injected into the DC microgrid to power the grid; c H For fuel prices; P H and PH,max These are the heat production power and maximum heat production power of the fuel cell cogeneration unit, respectively; P e,i For node i, the electrical load; P h,j For node j, the heat load is P. L,max This represents the maximum electro-thermal power of the fuel cell cogeneration unit.

[0114] Therefore, the method in this embodiment, by deeply decoupling the physical model from the underlying computing engine, allows engineers to focus on the design of the topology and related equations of the multi-energy system without needing to delve into the complex details of algorithm implementation. Based on the modular tensor flow configuration function provided by the industrial-grade configuration platform, users can easily build complex computational chains. This approach significantly reduces the difficulty and technical threshold of algorithm development, simplifies subsequent maintenance, and improves the efficiency of the entire system design and management. Automatic differentiation technology is used for matrix differentiation, combined with the hierarchical parallel processing mechanism based on directed acyclic graphs (DAGs) in the configuration platform, thereby significantly shortening the iteration time required for solving simultaneous nonlinear equations. This optimization not only accelerates the calculation speed but also enhances computational efficiency, meeting the real-time analysis needs of practical applications. Specifically, when facing complex multi-energy systems, this method can quickly provide accurate results, providing strong data support for decision support. Using tensors to represent state variables, combined with graphical configuration methods, ensures that when the system needs expansion or equipment changes, only the dimensions of the input tensors or local node connections need to be adjusted to complete the corresponding modifications, without the need for a complete reconstruction. This approach not only greatly simplifies the system update process but also effectively ensures the continuity and stability of multi-functional system operation monitoring. Therefore, even with changes in the system architecture, it can ensure efficient and stable operation, enhancing the system's flexibility and adaptability.

[0115] In summary, the method in this embodiment optimizes the entire process from algorithm development to system operation and maintenance through innovative design of the computing model. This not only improves work efficiency but also enhances the reliability and adaptability of the system, providing a brand-new solution for the intelligent management and control of multi-functional systems.

[0116] To verify the effectiveness and superiority of the multi-energy flow computation optimization model proposed in this embodiment, a simulation analysis was conducted based on a multi-energy system example consisting of a DC microgrid and a heating network in a real-world case. The design of the example followed the principles of typicality, reproducibility, and rationality, and a solution and visualization program was developed based on an industrial-grade configuration platform.

[0117] This example includes the following elements: This embodiment uses a multi-energy system composed of a DC microgrid and a heating network as an example for illustration. Its topology is as follows: Figure 4 and5 As shown in the figure, the system comprises 14 DC microgrid nodes and 9 heating network nodes. Node 0 is the connection point between the microgrid and the external power grid, and node 13 is the node where the fuel cell cogeneration unit is located. This node acts as an electro-thermal coupling hub, connecting the two sub-networks.

[0118] Equipment parameters: The data input to the data source node are shown in Tables 1 and 2, where NodeId, Is, Vs, ms and Tb are the node number, injected current, voltage, heat load flow and regeneration temperature, respectively, and SourceNode, TargetNode, Rk, L and D are the first node number, last node number, resistance, length and diameter of the branch, respectively.

[0119] Table 1. Node Data Tensors of Multi-Energy System

[0120] Table 2. Branch Data Tensors of Multi-Energy System

[0121] Based on the proposed model configuration, Figure 6 The corresponding tensor flow design is illustrated. In this design, nodes 1 through 4 are designated as data source nodes, with their data source type set to measurement points to acquire information on system load and energy prices. Node 5 generates the tensor corresponding to the objective function, while node 6 generates the tensor corresponding to the constraints. Node 7 is configured for mixed-integer linear programming (MILP) to solve the entire optimization model. Notably, this tensor flow triggering mechanism is data source-driven, meaning that optimization calculations are automatically triggered when system load or energy prices change.

[0122] Real-time operation of the above tensor flow for optimized control of the multi-energy system yields the following results: Figure 7As shown in the example, in the simulation settings, the electricity price is 0.8373 yuan / kWh during periods 0 to 55, and decreases to 0.3343 yuan / kWh during periods 56 to 95, while the fuel price remains constant at 0.465 yuan / kWh. The optimization results show that during periods with higher electricity prices (0 to 55), because the electricity price is higher than the fuel price, the fuel cell outputs its maximum power of 80kW, and the load of the thermal system is entirely supplied by the fuel cell. During periods with lower electricity prices (56 to 95), the system prioritizes grid power to meet energy demand. However, limited by the 40kW upper limit of the electricity-to-heat conversion power of the cogeneration unit, the fuel cell still needs to maintain a portion of its heating power output during periods 56 to 73; after period 74, the heating power of the fuel cell drops to zero, and the heat load is entirely supplied by the electricity-to-heat conversion method. These optimization results are consistent with the expectations of the economic analysis, demonstrating the effectiveness of the proposed method in multi-energy optimization scheduling scenarios.

[0123] also, Figure 8 The invention showcases the visualization and analysis capabilities of multi-energy flow computation and multi-energy optimization scheduling results, further highlighting its practical value in engineering applications. By adopting a modular configuration approach for tensor flow nodes and edges, engineers can quickly implement complex multi-energy flow computation and optimization scheduling algorithms, much like "building blocks," significantly reducing the time and maintenance costs required for traditional custom programming. This method not only improves development efficiency but also makes the system easier to maintain and expand, providing strong support for the intelligent management of multi-energy systems. In summary, this invention, through innovative design of the computational model, optimizes the entire process from algorithm development to system operation and maintenance, significantly improving work efficiency and system reliability, and providing a new solution for the efficient operation of multi-energy systems.

[0124] This embodiment aims to address several key issues in traditional multi-energy system energy flow calculations: deep coupling between the physical model and the solution engine, low efficiency in solving nonlinear equations, and poor adaptability to topology changes. To solve these problems, we propose an optimized solution architecture based on an industrial-grade configuration platform, which integrates tensor data structures and directed acyclic computation graphs. First, a multi-energy coupled physical model is established based on Kirchhoff's laws and thermodynamic equations, and the state data of this model is abstracted into a high-order tensor structure. This representation not only accurately describes the complex energy network state but also provides a flexible data manipulation interface. Second, a multi-mechanism driven tensor flow computation graph is constructed on the configuration platform. Through modular configuration, complete decoupling of the computation engine and problem modeling is achieved, meaning that users can freely choose or replace the computation engine without affecting the model definition process, and vice versa. This design greatly improves the system's flexibility and scalability. The core innovation lies in the introduction of automatic differentiation and matrix analysis techniques, automatically generating Jacobian and Hessian matrices for the nonlinear coupling parts in energy flow calculations. These matrices are crucial for accelerating the iterative convergence process of open-source or commercial solvers because they provide key information about the rate of change of the function, thus helping the solver find solutions more quickly and accurately. Ultimately, this method can significantly reduce the modeling difficulty of complex energy networks while maintaining the accuracy of multi-energy flow calculations, and improve the computational efficiency of handling multi-source heterogeneous data interactions and solving nonlinear equations. Therefore, this invention provides a powerful tool that enables engineers and technicians to optimize and control multi-energy systems more efficiently, while also promoting the development of intelligent energy management systems. This not only simplifies the development process and reduces maintenance costs, but also enhances the system's adaptability to future changes.

[0125] The aforementioned optimized implementation method for multi-energy flow computation based on an industrial-grade configuration platform effectively reduces model coupling and significantly improves nonlinear solution efficiency by implementing an optimized multi-energy flow computation method on an industrial-grade configuration platform. Specifically, firstly, a unified power grid and heating network computation model is established based on the transmission laws of different energy media in a multi-energy system, and power balance constraints are established to form a set of equations. Then, these model parameters are mapped into tensor data structures, constructing a directed acyclic tensor flow computation graph composed of multiple nodes, thus completely decoupling the computation engine from problem modeling. Next, solver interfaces are configured in the solution nodes of the computation graph, and standardized descriptions and required matrices are generated through automatic differentiation technology, thereby accelerating the solution process. Finally, the optimized computation graph is run using a data source or event-driven mechanism to achieve efficient sorting, scheduling, and parallel execution of computation nodes. The results can be easily mapped to visualization components, databases, or control units, which not only significantly enhances the computational efficiency of multi-energy system state analysis but also greatly increases configuration flexibility, providing a solid foundation for achieving more intelligent and efficient multi-energy system management.

[0126] Figure 9 This is a schematic block diagram of a multi-energy flow computing optimization implementation system 300 based on an industrial-grade configuration platform provided in an embodiment of the present invention. Figure 9 As shown, corresponding to the above-described multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform, this invention also provides a multi-energy flow computing optimization implementation system 300 based on an industrial-grade configuration platform. This multi-energy flow computing optimization implementation system 300 includes a unit for executing the above-described multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform, and the system can be configured in a server. Specifically, please refer to... Figure 9 The multi-energy flow computing optimization implementation system 300 based on an industrial-grade configuration platform includes a construction unit 301, a transformation unit 302, an optimization unit 303, and a solution mapping unit 304.

[0127] The construction unit 301 is used to establish calculation models of the power grid and heating network based on the transmission laws of different energy media in the multi-energy system, and to establish power balance constraints of the coupling nodes, forming a unified and computable set of equations and constraints to obtain a multi-energy flow coupled physical model. The conversion unit 302 is used to map all parameters and state variables in the multi-energy flow coupled physical model into tensor data with a unified structure, and to configure a directed acyclic tensor flow computation graph composed of multiple nodes. The optimization unit 303 is used to configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and to generate standardized descriptions and required matrices for the equations and tensor data through automatic differentiation technology to obtain an optimized directed acyclic tensor flow computation graph. The solution mapping unit 304 is used to run the optimized directed acyclic tensor flow computation graph using a data source or event-driven mechanism, sort and schedule the calculation nodes to achieve parallel execution, and then map the solution results to a visualization component, database, or control unit.

[0128] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned multi-energy flow computing optimization implementation system 300 and its various units based on the industrial-grade configuration platform can be referred to the corresponding descriptions in the aforementioned method embodiments. For the sake of convenience and brevity, these details will not be repeated here.

[0129] The aforementioned multi-energy flow computing optimization implementation system 300 based on an industrial-grade configuration platform can be implemented as a computer program, which can be used in, for example... Figure 10 It runs on the computer device shown.

[0130] Please see Figure 10 , Figure 10 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a server, wherein the server can be a standalone server or a server cluster composed of multiple servers.

[0131] See Figure 10 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0132] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform.

[0133] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0134] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform.

[0135] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0136] The processor 502 is used to run the computer program 5032 stored in the memory to implement all the steps of the multi-energy flow computing optimization implementation method based on the industrial-grade configuration platform.

[0137] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0138] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0139] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein when executed by a processor, the computer program causes the processor to perform all the steps of the multi-energy flow computing optimization implementation method based on an industrial-grade configuration platform.

[0140] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0141] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0142] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of each unit is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0143] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the system of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0144] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0145] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for optimizing multi-energy flow computation based on an industrial-grade configuration platform, characterized in that, include: Based on the transmission laws of different energy media in multi-energy systems, calculation models of power grid and heating network are established respectively, and power balance constraints of coupling nodes are established to form a unified and computable set of equations and constraints, so as to obtain a multi-energy flow coupling physical model. All parameters and state variables in the multi-energy flow coupled physical model are mapped to tensor data with a unified structure, and a directed acyclic tensor flow computation graph composed of multiple nodes is configured. Configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and generate standardized descriptions and required matrices for the equation set and tensor data through automatic differentiation techniques to obtain an optimized directed acyclic tensor flow computation graph. The optimized directed acyclic tensor flow computation graph is run using a data source or event-driven mechanism. The computation nodes are sorted and scheduled to achieve parallel execution. The solution results are then mapped to a visualization component, database, or control unit.

2. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 1, characterized in that, Based on the transmission patterns of different energy media in a multi-energy system, calculation models for the power grid and heating network are established respectively, and power balance constraints at coupled nodes are defined to form a unified and computable set of equations and constraints, thereby obtaining a multi-energy flow coupled physical model, including: Based on the physical laws of power transmission and Kirchhoff's laws, a power grid calculation model including nodal voltage equations is constructed for the electrical network within a multi-energy system. Based on the principle of mass and energy conservation, a hydraulic and thermal equation model is developed for the heating network to obtain a calculation model for the heating network. Within a multi-energy system, different energy networks are connected through energy conversion devices. Based on the input and output characteristics of the energy conversion devices, power balance constraints for the coupling nodes are established. The power grid calculation model and the heating network calculation model, together with the power balance constraints of the coupled nodes, are organized into a comprehensive set of equations and constraints to obtain a multi-energy flow coupled physical model.

3. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 1, characterized in that, The process of mapping all parameters and state variables in the multi-energy flow coupled physical model to tensor data with a unified structure, and configuring a directed acyclic tensor flow computation graph composed of various nodes, includes: Determine all key parameters and state variables that need to be solved in the multi-energy flow coupled physical model, and define the roles of key parameters and state variables in the multi-energy flow model to obtain the definition results; Based on dimensional planning, the parameters and state variables in the defined results are mapped to tensor data structures, and combined with topological data to reconstruct tensor groups to obtain tensor data; Based on the computational dependencies between the variables of the tensor data, a directed acyclic tensor flow computation graph consisting of data acquisition nodes, matrix operation nodes, constraint generation nodes, and solution nodes is configured in the industrial-grade configuration platform, and the reading, splicing, transposing, aggregation, and distribution paths of tensors are defined by connecting the nodes.

4. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 3, characterized in that, The process of mapping the parameters and state variables in the defined results to tensor data structures based on dimensional planning, and reconstructing them into tensor groups by combining topological data, to obtain tensor data, includes: By extending data types using multidimensional arrays of tensors, topological and state data such as branch correlation matrices, branch conductivity matrices, and heat medium flow rate vectors in multi-energy systems are reconstructed into tensor groups suitable for parallel matrix operations at the platform's underlying layer, combined with name strings that identify the dimension length. Dimension labels, index mapping relationships, and data type identifiers are configured for each tensor to support automatic addressing, slice calling, and inter-node transfer within the configuration platform.

5. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 4, characterized in that, The node types of the directed acyclic tensor flow computation graph include data acquisition nodes, tensor conventional computation nodes, and solution nodes for nonlinear equation systems or mixed integer programming; the edges of the directed acyclic tensor flow computation graph are used to define the tensor data transfer paths between the first and last nodes and the aggregation and transpose preprocessing operations.

6. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 1, characterized in that, The process of configuring linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and generating standardized descriptions and required matrices for the equation set and tensor data using automatic differentiation techniques to obtain an optimized directed acyclic tensor flow computation graph, includes: Configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph; The system of equations and the tensor data are taken as input and passed to the parsing engine to perform lexical analysis on the formula string, identify variables and match their indices, and form an expression tree; Based on the expression tree, the first and second derivatives required in the solution process are automatically generated using automatic differentiation techniques. The first and second derivatives, along with the constraint boundary conditions, initial values, and convergence criteria, are encapsulated into a standardized input format required by the solver to obtain the optimized directed acyclic tensor flow computation graph.

7. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 6, characterized in that, The expression includes a standardized description of the objective function, constraint functions, and upper and lower bounds of the variables.

8. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 1, characterized in that, The process of running the optimized directed acyclic tensor flow computation graph using a data source or event-driven mechanism, sorting and scheduling computation nodes to achieve parallel execution, and then mapping the solution results to a visualization component, database, or control unit includes: Multiple triggering mechanisms are set up to activate the corresponding computation graph branch in the optimized directed acyclic tensor flow computation graph when real-time data updates or system parameter changes are detected. The activated computation graph branches automatically sort and schedule computation nodes according to the topology hierarchy, and allocate independent threads to subgraphs without data dependencies to achieve parallel processing, so as to perform operations on each computation node. Among them, each computation node performs matrix transformation, nonlinear solution and other operations in the scheduling order to complete the solution of the multi-energy flow model and output energy allocation results and equipment control commands. The energy allocation results, computing node status variables, and device control commands are synchronously mapped to a visual interface for display, stored in a historical database for analysis, and sent to the underlying control unit for physical device regulation.

9. The multi-energy flow computation optimization implementation method based on an industrial-grade configuration platform according to claim 8, characterized in that, The various triggering mechanisms include time-driven, data-driven, and event-driven mechanisms configured according to operational requirements. Among them, time-driven mechanisms are used to perform multi-energy flow state calculations and optimization scheduling according to a preset cycle; data-driven mechanisms are used to trigger calculations when changes in real-time measurement data, boundary conditions, or price parameters reach a set threshold; and event-driven mechanisms are used to trigger the calculation process when equipment starts or stops, switches between states, fault alarms are triggered, or control commands are issued.

10. A multi-energy flow computing optimization implementation system based on an industrial-grade configuration platform, characterized in that, include: The building unit is used to establish calculation models of power grid and heating network based on the transmission laws of different energy media in multi-energy systems, and to establish power balance constraints of coupling nodes, forming a unified and computable set of equations and constraints to obtain a multi-energy flow coupling physical model. The conversion unit is used to map all parameters and state variables in the multi-energy flow coupled physical model into tensor data with a unified structure, and to configure a directed acyclic tensor flow computation graph composed of multiple nodes. An optimization unit is used to configure linear or nonlinear solver interfaces in the solution nodes of the directed acyclic tensor flow computation graph, and to generate a standardized description and required matrix for the equation set and the tensor data through automatic differentiation technology to obtain an optimized directed acyclic tensor flow computation graph. The solution mapping unit is used to run the optimized directed acyclic tensor flow computation graph using a data source or event-driven mechanism, sort and schedule the computation nodes to achieve parallel execution, and then map the solution results to a visualization component, database or control unit.

Images