Multi-disciplinary design simulation optimization parallel computing system and method

Abstract

The invention discloses a multidisciplinary design simulation optimization parallel computing system and a multidisciplinary design simulation optimization parallel computing method. The multidisciplinary design simulation optimization parallel computing system comprises a local workstation integrating a plurality of optimization layers and a cloud super computer which is in communication interaction with each optimization layer. Wherein after the local workstation initiates a multidisciplinary design optimization task, a corresponding parallel optimization strategy is allocated to each optimization layer based on the initiated multidisciplinary design optimization task, so that each optimization layer establishes a corresponding optimization calculation task based on the allocated parallel optimization strategy, and the optimization calculation task is uploaded to the cloud super computer for solving. After receiving the optimization calculation tasks uploaded by the optimization layers, the cloud super computer allocates corresponding simulation parallel strategies and calculation resources to each optimization calculation task, and then solves the optimization calculation task of each optimization layer based on the allocated simulation parallel strategies and calculation resources; and the solving result of each optimization calculation task is issued to the corresponding optimization layer for local iterative optimization until the whole multidisciplinary design optimization task is completed.

Description

Technical Field

[0001] This invention relates to the field of simulation optimization, and more specifically to a multidisciplinary design simulation optimization parallel computing system and method. Background Technology

[0002] With the emergence and development of computer hardware technology, the scale of various complex engineering and product problems solved using its high-efficiency computing performance in combination with multidisciplinary and multi-objective optimization and numerical simulation is becoming increasingly larger, resulting in excessively long solution times that cannot meet the requirements of the R&D cycle for engineering and product problems. In order to shorten the optimization and simulation solution time of large-scale engineering and product systems, it is necessary to improve the optimization and simulation process with the help of parallel computing technology to enhance its parallel computing capabilities and solution efficiency.

[0003] The key to optimizing parallel computing is to employ load balancing strategies to allocate appropriate computing resources to the optimization task. Compared to numerical simulation, the optimization algorithm itself has a relatively small computational load; therefore, improving the efficiency of parallel computing for simulation optimization hinges on improving the computational efficiency of numerical simulation.

[0004] In parallel computing analysis of numerical simulations, the portion that consumes the largest proportion of time is often the solution of linear equation systems. Scholars at home and abroad generally classify parallel solution algorithms for linear equations into two categories: direct methods and iterative methods. The parallel computer systems they are adapted to are mainly homogeneous supercomputers with pure CPUs, and heterogeneous supercomputers with multi-core CPUs and GPUs or many-core processors (MICs).

[0005] When using direct methods for parallel solutions, the exact solution to the system is typically obtained within the expected computational steps through sorting, triangular decomposition, and back substitution. Since the computational order and computational load are deterministic, the overall parallel solution time can be predicted with relatively high accuracy. However, the computation process often alters the sparsity of the coefficient matrix, resulting in significant memory consumption, and the computational load increases rapidly with the size of the coefficient matrix, making its parallel scalability inferior to iterative methods.

[0006] Iterative parallel solutions improve the solution through multiple iterations, attempting to converge to an exact solution. The overall solution process does not change the sparsity of the coefficient matrix, resulting in less communication overhead and thus lower memory requirements, while also exhibiting good parallel scalability. However, the convergence of the overall solution is limited by the condition number of the coefficient matrix; if the condition number is too large, the overall convergence is poor.

[0007] Therefore, the parallel algorithms provided by current optimization software have certain limitations and cannot meet the computational efficiency requirements of large-scale optimization problems. Furthermore, in application areas such as aero-engines, multi-disciplinary and multi-objective hierarchical optimization is often required, and the allocation of computational resources between and within each sub-optimization task cannot achieve load balancing of supercomputer hardware and software resources. Summary of the Invention

[0008] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form to prepare for the more detailed descriptions that follow.

[0009] The purpose of this invention is to solve the above-mentioned problems and provide a multidisciplinary design simulation optimization parallel computing system and method. Multiple optimization layers are integrated on a local workstation, and then the initiated multidisciplinary design optimization tasks are distributed to each local optimization layer for parallel computing according to discipline and physical framework. At the same time, each optimization layer uploads the computationally intensive optimization tasks to a cloud supercomputer for parallel computing, thereby improving the simulation optimization efficiency.

[0010] The technical solution of this invention is as follows:

[0011] This invention provides a multidisciplinary design simulation optimization parallel computing system, integrating local workstations with multiple optimization layers and a cloud-based supercomputer that communicates and interacts with each optimization layer; wherein,

[0012] The local workstation deploys an optimization layer parallel strategy allocation module and an optimization parallel strategy module. After the local workstation initiates a multidisciplinary design optimization task, it allocates the corresponding parallel optimization strategy from the optimization parallel strategy module to the corresponding optimization layer based on the initiated multidisciplinary design optimization task. This allows each optimization layer to establish its corresponding optimization calculation task based on the allocated parallel optimization strategy and upload it to the cloud supercomputer for solving.

[0013] The cloud-based supercomputer deploys a simulation parallel strategy module, a resource scheduling module, and a parallel algorithm library. After receiving optimization computation tasks uploaded from each optimization layer, the cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each optimization computation task through the simulation parallel strategy module and the resource scheduling module. Then, based on the allocated simulation parallel strategies and computing resources, it calls parallel algorithms from the parallel algorithm library to solve the optimization computation tasks of each optimization layer, and sends the solution results of each optimization computation task to the corresponding optimization layer for local iterative optimization until the entire multidisciplinary design optimization task is completed.

[0014] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, the optimization layer is divided into a whole-system layer, a component layer, a part layer, and a discipline layer. After the local workstation initiates a multidisciplinary design optimization task, it assigns corresponding optimization parallel strategies to the integrated whole-system layer, component layer, part layer, and discipline layer. The parallel optimization strategies include parallel allocation strategies for the whole-system layer, component layer, part layer, and discipline layer. After each optimization layer obtains the assigned parallel optimization strategy, it establishes a corresponding optimization computing task based on the assigned parallel optimization strategy, and then uploads the established optimization computing task to a cloud supercomputer for solving.

[0015] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, after the cloud supercomputer receives the optimization computing tasks uploaded by each optimization layer, it allocates the corresponding simulation parallel strategy to the corresponding optimization computing task from the simulation parallel strategy module according to the discipline module corresponding to each optimization computing task. Then, based on the allocated simulation parallel strategy, the corresponding optimization computing task is decomposed into multiple task packages that can be operated on the solution nodes of the cloud supercomputer and solved, thereby completing the solution of the optimization computing tasks uploaded by each optimization layer.

[0016] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, the discipline modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies according to the discipline modules corresponding to each optimization calculation task. Each discipline module then decomposes the corresponding optimization calculation task into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solves them based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies.

[0017] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, the cloud supercomputer decomposes the optimization computing tasks uploaded by each optimization layer into multiple task packages, allocates corresponding computing resources to each task package through a resource scheduling module, and then solves each task package based on the allocated computing resources. The resource scheduling module includes an MPI accelerator, a computational load statistical analysis program, and a load balancing resource scheduler. When allocating computing resources to each task package, the resource scheduling module first performs MPI compilation and transformation on each task package through the MPI accelerator, then distributes the compiled and transformed task packages to each solution node, estimates the computational load of each transformed task package through the computational load statistical analysis program, and finally allocates corresponding computing resources to each task package through the load balancing resource scheduler, thereby utilizing the allocated computing resources to solve the task packages on each solution node.

[0018] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, after the cloud supercomputer completes the resource allocation of each task package, it calls the parallel algorithms in the parallel algorithm library to solve the task packages at each solution point; wherein, the parallel algorithms include the direct method for parallel computation of linear equations, the iterative method for parallel computation of linear equations, the domain decomposition parallel computation method, and the hierarchical parallel computation method.

[0019] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, after the cloud supercomputer completes the solution of the task package corresponding to each optimization calculation task, it sends the solution results of each task package to the optimization layer on the corresponding local computer; wherein, after receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results of each optimization layer.

[0020] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, when each optimization layer performs local iterative optimization, the optimization result of the previous layer is also introduced as an input parameter to solve the optimization result, and the optimization result of the next layer is used as a constraint condition to solve the optimization result of the current layer.

[0021] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, the multidisciplinary design simulation optimization parallel computing system further includes a data transmission module for data transmission between optimization layers of the local computer, between each optimization layer and the cloud supercomputer, and between each solution node of the cloud supercomputer and within the solution node; wherein, the data transmission module includes: a data transmission volume balancing scheduler, an inter-node transmission controller, an intra-node transmission controller, and a local / supercomputer upload / download controller.

[0022] According to an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention, the multidisciplinary design simulation optimization parallel computing system further includes a high-performance auxiliary module for providing supercomputing resources required for simulation optimization parallel computing for cloud supercomputers; wherein, the high-performance auxiliary module includes: a supercomputing high-speed computing management unit, a supercomputing high-speed computing storage unit, a supercomputing high-speed computing unit, and a supercomputing high-speed network unit.

[0023] This invention provides a multidisciplinary design simulation optimization parallel computing method, comprising the following steps:

[0024] Step S1: Initiate a multidisciplinary design optimization task based on a local workstation with multiple optimization layers deployed;

[0025] Step S2: Assign corresponding parallel optimization strategies to each optimization layer of the local workstation based on the initiated multidisciplinary design optimization task;

[0026] Step S3: Each optimization layer establishes a corresponding optimization computation task based on the assigned parallel optimization strategy and uploads it to the cloud supercomputer for solving;

[0027] Step S4: The cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each received optimization computing task;

[0028] Step S5: The cloud-based supercomputer uses parallel algorithms from the parallel algorithm library to solve the optimization tasks of each optimization layer based on the allocated simulation parallel strategy and computing resources, and obtains the solution results of each optimization task.

[0029] Step S5: The cloud-based supercomputer distributes the solution results of each optimization calculation task to the corresponding optimization layers for local iterative optimization until the entire multidisciplinary design optimization task is completed.

[0030] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, the optimization layer is divided into a whole machine layer, a component layer, a part layer, and a discipline layer. After the local workstation initiates a multidisciplinary design optimization task, it assigns corresponding optimization parallel strategies to the integrated whole machine layer, component layer, part layer, and discipline layer. After each optimization layer obtains the assigned optimization parallel strategy, it establishes a corresponding optimization computing task based on the assigned parallel optimization strategy, and then uploads the established optimization computing task to the cloud supercomputer for solving.

[0031] According to one embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, after receiving the optimization computing tasks uploaded by each optimization layer, the cloud supercomputer allocates a corresponding simulation parallel strategy to each optimization computing task according to the subject module corresponding to each optimization computing task. Then, based on the allocated simulation parallel strategy, the corresponding optimization computing task is decomposed into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solved, thereby completing the solution of the optimization computing tasks uploaded by each optimization layer.

[0032] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, the discipline modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies according to the discipline modules corresponding to each optimization computing task. Then, based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies, the corresponding optimization computing task is decomposed into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solved.

[0033] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, the cloud supercomputer decomposes the optimization computing tasks uploaded from each optimization layer into multiple task packages, allocates corresponding computing resources to each task package, and then solves each task package based on the allocated computing resources. When allocating computing resources to each task package, the multidisciplinary design simulation optimization parallel computing method first performs MPI compilation and transformation on each task package through an MPI accelerator, then distributes the compiled and transformed task packages to each solver node, estimates the computational workload of each transformed task package through a computational workload statistical analysis program, and finally allocates corresponding computing resources to each task package through a load balancing resource scheduler, thereby using the allocated computing resources to solve the task packages on each solver node.

[0034] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, after the cloud supercomputer completes the resource allocation of each task package, it calls the parallel algorithms in the parallel algorithm library to solve the task packages at each solution point; wherein, the parallel algorithms include the direct method for parallel computing of linear equations, the iterative method for parallel computing of linear equations, the domain decomposition parallel computing method, and the hierarchical parallel computing method.

[0035] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, after the cloud supercomputer completes the solution of the task package corresponding to each optimization calculation task, it sends the solution results of each task package to the optimization layer on the corresponding local computer; wherein, after receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results of each optimization layer.

[0036] According to an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention, when each optimization layer performs local iterative optimization, the optimization result of the previous layer is also introduced as an input parameter to solve the optimization result, and the optimization result of the next layer is used as a constraint condition to solve the optimization result of the current layer.

[0037] The present invention also provides a computer-readable medium storing computer program code that, when executed by a processor, implements the method described above.

[0038] The present invention also provides a multidisciplinary design simulation optimization parallel computing device, comprising:

[0039] Memory, used to store instructions that can be executed by a processor; and

[0040] A processor for executing the instructions to implement the method described above.

[0041] Compared with existing technologies, this invention offers the following advantages: For simulation optimization of multidisciplinary designs, this invention integrates multiple optimization layers on a local workstation, each communicating with a cloud-based supercomputer. The simulation optimization task of the multidisciplinary design to be optimized is distributed to each local optimization layer for parallel computation according to discipline and physical framework. Simultaneously, each optimization layer uploads its computationally intensive optimization tasks to the cloud-based supercomputer for parallel computation, thereby improving simulation optimization efficiency. Compared with existing technologies, this invention uploads complex and computationally intensive optimization tasks to the cloud-based supercomputer, applying parallel algorithms, parallel strategies, and resource scheduling to the simulation optimization process. While ensuring numerical simulation accuracy and optimization effect, it utilizes the massive computing power of the cloud-based supercomputer for parallel computation, significantly improving simulation optimization computation efficiency, saving computational costs, shortening product development time, and enabling high-precision solutions to ultra-large-scale optimization problems. Furthermore, based on computational load balancing, this invention also calculates and detects real-time load balancing of data transmission within each computing node, between nodes, and at each level, thereby improving data exchange efficiency and further enhancing simulation optimization computation efficiency. Attached Figure Description

[0042] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0043] Figure 1 This is a system architecture diagram illustrating an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention.

[0044] Figure 2 This is a data flow diagram illustrating an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention.

[0045] Figure 3 This is a logic framework diagram illustrating an embodiment of the MDO optimization of an aero-engine according to the present invention.

[0046] Figure 4 This is a flowchart illustrating one embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention. Detailed Implementation

[0047] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this application. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0048] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0049] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0050] In detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure will be partially enlarged and not to scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of the present invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0051] In the description of this application, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is usually based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this application and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this application; the directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0052] This document discloses an embodiment of a multidisciplinary design simulation optimization parallel computing system (hereinafter sometimes referred to as an optimization parallel computing system). Figure 1 This is a system architecture diagram illustrating an embodiment of the multidisciplinary design simulation optimization parallel computing system of the present invention. Figure 1 As shown, in this embodiment, the optimized parallel computing system includes a local workstation integrating multiple optimization layers and a cloud-based supercomputer that communicates and interacts with each optimization layer. The local workstation deploys an optimization layer parallel strategy allocation module and an optimization parallel strategy module. After the local workstation initiates a multidisciplinary design optimization task, it allocates corresponding parallel optimization strategies from the optimization parallel strategy module to the corresponding optimization layers through the optimization layer parallel strategy allocation module. This allows each optimization layer to establish its own optimization computation task based on the allocated parallel optimization strategy and upload it to the cloud-based supercomputer for solving.

[0053] The cloud-based supercomputer deploys a simulation parallel strategy module, a resource scheduling module, and a parallel algorithm library. Upon receiving optimization tasks uploaded from each optimization layer, the cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each task using the simulation parallel strategy module and the resource scheduling module. Then, based on the allocated simulation parallel strategies and computing resources, it calls parallel algorithms from the parallel algorithm library to solve the optimization tasks at each optimization layer. The results of each optimization task are then distributed to the corresponding optimization layers for local iterative optimization until the entire multidisciplinary design optimization task is completed.

[0054] Specifically, in this embodiment, the optimization layer is divided into a system layer, a component layer, a part layer, and a discipline layer. After the local workstation initiates a multidisciplinary design optimization task, corresponding parallel optimization strategies are assigned to the integrated system layer, component layer, part layer, and discipline layer. These parallel optimization strategies include parallel allocation strategies for the system layer, component layer, part layer, and discipline layer. These strategies are then assigned to the corresponding system layer, component layer, part layer, and discipline layer to achieve multidisciplinary and multi-objective coordination within and between each layer.

[0055] After each optimization layer receives its assigned parallel optimization strategy, it establishes corresponding optimization computation tasks based on these strategies for parallel simulation and solution. During the solution process, relying on multidisciplinary optimization software, each optimization layer performs performance optimization or one-dimensional simulation with relatively small computational loads on its local workstation, allocating and solving parallel resources accordingly. For computationally intensive optimization tasks, such as two-dimensional / three-dimensional numerical simulations, these tasks are uploaded to a cloud-based supercomputer for parallel resource allocation and solution, avoiding the need to deploy the entire optimization environment on the supercomputer. The overall system layer, component layer, part layer, and discipline layer need to be solved sequentially from top to bottom, but parallel computation can be performed between components within the component layer, between parts within the part layer, and between disciplines within the discipline layer.

[0056] Specifically, in this embodiment, when each optimization layer calls upon the cloud-based supercomputer to perform parallel simulations of computationally intensive optimization tasks, it uses the supercomputer's high-speed computing network to upload parallel strategies and simulation input files by discipline. After receiving the optimization tasks uploaded by each optimization layer, the cloud-based supercomputer allocates corresponding simulation parallel strategies from the simulation parallel strategy module to the corresponding optimization task based on the discipline module. Then, based on the allocated simulation parallel strategies, it decomposes the corresponding optimization task into multiple task packages that can be computed on the cloud-based supercomputer's solution nodes and solves them, thereby completing the solution of the optimization tasks uploaded by each optimization layer. Figure 2 This is a data flow diagram illustrating an embodiment of parallel simulation optimization of the entire aero-engine MDO (Main Engine Design). Figure 3 This is a logical framework diagram illustrating an embodiment of the parallel simulation optimization of MDO for aero-engines according to the present invention. The following describes this embodiment in more detail using the parallel simulation optimization of MDO for aero-engines as an example.

[0057] like Figure 2 and Figure 3 As shown, the MDO (Model-Oriented Optimization) of an aero-engine comprises a four-layered disciplinary optimization logic architecture, including the entire aero-engine, unit components such as compressor / turbine / combustion chamber, parts such as fan casing / fan blades / compressor casing, and strength / aerodynamics. The optimization results of the previous layer serve as the parameter input for the next layer, and the optimization results of the next layer serve as the criteria for satisfying the optimization constraints of the previous layer. Its disciplinary modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud-based supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies based on the disciplinary modules corresponding to each optimization calculation task. Then, based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies, the corresponding optimization calculation task is decomposed into multiple task packages that can be computed on the cloud-based supercomputer's solution nodes and solved.

[0058] The cloud-based supercomputer decomposes the optimization computing tasks uploaded from each optimization layer into multiple task packages, allocates corresponding computing resources to each task package through the resource scheduling module, and then solves each task package based on the allocated computing resources.

[0059] Specifically, in this embodiment, the resource scheduling module includes an MPI accelerator, a computational load statistical analysis program, and a load balancing resource scheduler. When allocating computing resources to each task package, the resource scheduling module first performs MPI compilation and transformation on each task package using the MPI accelerator. It employs either a parallel algorithm using the iterative method for linear equations or a parallel algorithm using the direct method for linear equations to perform an equivalent transformation of the equations to improve parallel computing efficiency. After the MPI accelerator completes compilation, it distributes the transformed task packages to each solution node, where each solution node performs synchronous parallel computing. During processing, the computational load statistical analysis program can estimate the computational load of each transformed task package. Then, the load balancing resource scheduler allocates corresponding computing resources to each task package, thereby utilizing the allocated computing resources to solve the task packages on each solution node.

[0060] In allocating computing resources, parallel computing tasks are prioritized within heterogeneous core groups within a node. For computing tasks exceeding the computational capacity of a single core group, the allocation is extended across heterogeneous core groups within the node. For computing tasks exceeding the computational capacity of a single node, the allocation is further distributed to other nodes. These heterogeneous core groups include, but are not limited to, multi-core CPUs, GPUs, and many-core processors (MICs). This achieves equivalent accelerated solution of equations for each task package and a rational allocation of supercomputer hardware and software resources, thereby improving computational efficiency.

[0061] Furthermore, in this embodiment, during the iterative process, a computational workload statistical analysis program can be used to dynamically correct the estimated computational workload of each task package after transformation, obtaining dynamic change data on the computational workload, which is then used as input to the load balancer. This enables real-time updates to the computational workload, thereby dynamically allocating and adjusting the solution resources assigned to each computational task package.

[0062] In this embodiment, after the cloud-based supercomputer completes the resource allocation for each task package, each subject module begins to call parallel algorithms from the parallel algorithm library to solve the task packages at each solution point. These parallel algorithms include direct parallel computation methods for linear equation systems, iterative parallel computation methods for linear equation systems, domain decomposition parallel computation methods, and hierarchical parallel computation methods. Each module can call parallel acceleration methods such as domain decomposition and hierarchical methods from the parallel algorithm library based on the numerical simulation characteristics of its respective subject, or it can directly use parallel computation methods for solving linear equation systems.

[0063] When solving linear equations using the direct parallel computation method, three types of solution methods can be employed, including but not limited to variable-length solvers, sparse (dense) solvers, and multi-wavefront solvers, to obtain the exact solution of the system within the expected computation steps. When solving linear equations using the iterative parallel computation method, classical iterative methods can be employed, including but not limited to simple iterative methods, Jacobi iterative methods, Gauss-Seidel methods, over-relaxation iterative methods, Newton-Rafael methods, etc., or iterative methods based on Krylov subspaces, including minimum residual methods, conjugate gradient methods, and preconditional conjugate gradient methods. These methods improve the solution results through multiple iterations using sorting, triangular decomposition, and back-substitution to attempt convergence to the exact solution of the problem. Domain decomposition parallel computation methods are suitable for accelerating explicit dynamic solutions, while hierarchical parallel computation methods are suitable for accelerating static solutions.

[0064] In this embodiment, after the cloud-based supercomputer completes the solution of the task packages corresponding to each optimization computation task, it sends the solution results of each task package to the corresponding optimization layer on the local computer. Upon receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results for each optimization layer. When each optimization layer downloads the solution results from the cloud-based supercomputer for local iterative optimization, it also introduces the optimization results of the previous layer as input parameters to solve for the optimization results, and uses the optimization results of the next layer as constraints, thereby solving for the optimization results of its own layer.

[0065] Furthermore, in this embodiment, the simulation optimization parallel computing system also includes a data transmission module for data transmission between optimization layers on the local computer, between each optimization layer and the cloud supercomputer, and between and within each solution node of the cloud supercomputer. The data transmission module includes: a data transmission load balancing scheduler, an inter-node transmission controller, an intra-node transmission controller, and a local / supercomputer upload / download controller.

[0066] Meanwhile, to provide the supercomputing resources required for simulation optimization parallel computing to the cloud supercomputer, the simulation optimization parallel computing system also includes a high-performance auxiliary module, comprising a supercomputing high-speed computing management unit, a supercomputing high-speed computing storage unit, a supercomputing high-speed computing unit, and a supercomputing high-speed network unit. This built-in high-performance auxiliary module enables the cloud supercomputer to perform multidisciplinary optimization simulation parallel computing. Furthermore, the simulation optimization parallel computing system disclosed in this specification can be embedded in the multidisciplinary optimization software user interface as a sub-interface for invocation, or it can be used as an independent system, invoked by the multidisciplinary optimization software when parallel optimization is required.

[0067] This specification also discloses an embodiment of a multidisciplinary design simulation optimization parallel computing method. Figure 4This is a flowchart illustrating an embodiment of the multidisciplinary design simulation optimization parallel computing method of the present invention. Please refer to... Figure 4 The following is a detailed explanation of each step in the multidisciplinary design simulation optimization parallel computing method.

[0068] Step S1: Initiate a multidisciplinary design optimization task based on a local workstation with multiple optimization layers deployed.

[0069] Step S2: Assign corresponding parallel optimization strategies to each optimization layer of the local workstation based on the initiated multidisciplinary design optimization task.

[0070] In this embodiment, the optimization layer is divided into a system layer, a component layer, a part layer, and a discipline layer. After the local workstation initiates a multidisciplinary design optimization task, corresponding parallel optimization strategies are assigned to the integrated system layer, component layer, part layer, and discipline layer. These parallel optimization strategies include parallel allocation strategies for the system layer, component layer, part layer, and discipline layer. These strategies are then assigned to the corresponding system layer, component layer, part layer, and discipline layer to achieve multidisciplinary and multi-objective coordination within and between each layer.

[0071] Step S3: Each optimization layer establishes a corresponding optimization computation task based on the assigned parallel optimization strategy and uploads it to the cloud supercomputer for solving.

[0072] In this embodiment, after each optimization layer obtains its assigned parallel optimization strategy, it establishes corresponding optimization computation tasks based on the assigned parallel optimization strategy for parallel simulation and solution. During the solution process, relying on multidisciplinary optimization software, each optimization layer performs performance optimization or one-dimensional simulation with relatively small computational loads on its local workstation, allocating and solving parallel resources. If computationally intensive tasks are involved, such as two-dimensional / three-dimensional numerical simulations, these computationally intensive optimization computation tasks are uploaded to a cloud supercomputer for high-computation simulation parallel resource allocation and solution, thus avoiding the need to deploy the entire optimization environment on the supercomputer. The overall system layer, component layer, part layer, and discipline layer need to be solved sequentially from top to bottom, but parallel computation can be performed between components in the component layer, between parts in the part layer, and between disciplines in the discipline layer.

[0073] Step S4: The cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each received optimized computing task.

[0074] In this embodiment, when each optimization layer invokes a cloud-based supercomputer to perform parallel simulations of computationally intensive optimization tasks, it uses a high-speed computing network to upload parallel strategies and simulation input files by discipline. After receiving the optimization tasks uploaded by each optimization layer, the cloud-based supercomputer allocates corresponding simulation parallel strategies from the simulation parallel strategy module based on the discipline module corresponding to each optimization task. Then, based on the allocated simulation parallel strategies, it decomposes the corresponding optimization task into multiple task packages that can be computed on the cloud-based supercomputer's solution nodes and solves them, thereby completing the solution of the optimization tasks uploaded by each optimization layer. Figure 2 This is a data flow diagram illustrating an embodiment of parallel simulation optimization of the entire aero-engine MDO (Main Engine Design). Figure 3 This is a logical framework diagram illustrating an embodiment of the parallel simulation optimization of MDO for aero-engines according to the present invention. The following describes this embodiment in more detail using the parallel simulation optimization of MDO for aero-engines as an example.

[0075] like Figure 2 and Figure 3 As shown, the MDO (Model-Oriented Optimization) of an aero-engine comprises a four-layered disciplinary optimization logic architecture, including the entire aero-engine, unit components such as compressor / turbine / combustion chamber, parts such as fan casing / fan blades / compressor casing, and strength / aerodynamics. The optimization results of the previous layer serve as the parameter input for the next layer, and the optimization results of the next layer serve as the criteria for satisfying the optimization constraints of the previous layer. Its disciplinary modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud-based supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies based on the disciplinary modules corresponding to each optimization calculation task. Then, based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies, the corresponding optimization calculation task is decomposed into multiple task packages that can be computed on the cloud-based supercomputer's solution nodes and solved.

[0076] The cloud-based supercomputer decomposes the optimization computing tasks uploaded from each optimization layer into multiple task packages, allocates corresponding computing resources to each task package through the resource scheduling module, and then solves each task package based on the allocated computing resources.

[0077] Specifically, in this embodiment, the resource scheduling module includes an MPI accelerator, a computational load statistical analysis program, and a load balancing resource scheduler. When allocating computing resources to each task package, the resource scheduling module first performs MPI compilation and transformation on each task package using the MPI accelerator. It employs either a parallel algorithm using the iterative method for linear equations or a parallel algorithm using the direct method for linear equations to perform an equivalent transformation of the equations to improve parallel computing efficiency. After the MPI accelerator completes compilation, it distributes the transformed task packages to each solution node, where each solution node performs synchronous parallel computing. During processing, the computational load statistical analysis program can estimate the computational load of each transformed task package. Then, the load balancing resource scheduler allocates corresponding computing resources to each task package, thereby utilizing the allocated computing resources to solve the task packages on each solution node.

[0078] In allocating computing resources, parallel computing tasks are prioritized within heterogeneous core groups within a node. For computing tasks exceeding the computational capacity of a single core group, the allocation is extended across heterogeneous core groups within the node. For computing tasks exceeding the computational capacity of a single node, the allocation is further distributed to other nodes. These heterogeneous core groups include, but are not limited to, multi-core CPUs, GPUs, and many-core processors (MICs). This achieves equivalent accelerated solution of equations for each task package and a rational allocation of supercomputer hardware and software resources, thereby improving computational efficiency.

[0079] Furthermore, in this embodiment, during the iterative process, a computational workload statistical analysis program can be used to dynamically correct the estimated computational workload of each task package after transformation, obtaining dynamic change data on the computational workload, which is then used as input to the load balancer. This enables real-time updates to the computational workload, thereby dynamically allocating and adjusting the solution resources assigned to each computational task package.

[0080] Step S5: The cloud-based supercomputer uses parallel algorithms from the parallel algorithm library to solve the optimization tasks of each optimization layer based on the allocated simulation parallel strategy and computing resources, and obtains the solution results of each optimization task.

[0081] In this embodiment, after the cloud-based supercomputer completes the resource allocation for each task package, each subject module begins to call parallel algorithms from the parallel algorithm library to solve the task packages at each solution point. These parallel algorithms include direct parallel computation methods for linear equation systems, iterative parallel computation methods for linear equation systems, domain decomposition parallel computation methods, and hierarchical parallel computation methods. Each module can call parallel acceleration methods such as domain decomposition and hierarchical methods from the parallel algorithm library based on the numerical simulation characteristics of its respective subject, or it can directly use parallel computation methods for solving linear equation systems.

[0082] When solving linear equations using the direct parallel computation method, three types of solution methods can be employed, including but not limited to variable-length solvers, sparse (dense) solvers, and multi-wavefront solvers, to obtain the exact solution of the system within the expected computation steps. When solving linear equations using the iterative parallel computation method, classical iterative methods can be employed, including but not limited to simple iterative methods, Jacobi iterative methods, Gauss-Seidel methods, over-relaxation iterative methods, Newton-Rafael methods, etc., or iterative methods based on Krylov subspaces, including minimum residual methods, conjugate gradient methods, and preconditional conjugate gradient methods. These methods improve the solution results through multiple iterations using sorting, triangular decomposition, and back-substitution to attempt convergence to the exact solution of the problem. Domain decomposition parallel computation methods are suitable for accelerating explicit dynamic solutions, while hierarchical parallel computation methods are suitable for accelerating static solutions.

[0083] Step S6: The cloud-based supercomputer distributes the solution results of each optimization calculation task to the corresponding optimization layers for local iterative optimization until the entire multidisciplinary design optimization task is completed.

[0084] In this embodiment, after the cloud-based supercomputer completes the solution of the task packages corresponding to each optimization computation task, it sends the solution results of each task package to the corresponding optimization layer on the local computer. Upon receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results for each optimization layer. When each optimization layer downloads the solution results from the cloud-based supercomputer for local iterative optimization, it also introduces the optimization results of the previous layer as input parameters to solve for the optimization results, and uses the optimization results of the next layer as constraints, thereby solving for the optimization results of its own layer.

[0085] This specification also provides a computer-readable medium storing computer program code that, when executed by a processor, implements the multidisciplinary design, simulation, optimization, and parallel computing system as described above.

[0086] This specification also provides a multidisciplinary design simulation optimization parallel computing device, including a memory storing instructions executable by a processor, and a processor for executing instructions in the instruction memory to implement the multidisciplinary design simulation optimization parallel computing system as described above.

[0087] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0088] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0089] The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein can be implemented or performed using a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The general-purpose processor may be a microprocessor, but in alternatives, it may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors cooperating with a DSP core, or any other such configuration.

[0090] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of both. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor such that the processor can read and write information to / from the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside as discrete components in the user terminal.

[0091] In one or more exemplary embodiments, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functionality may be stored or transmitted as one or more instructions or code on or through a computer-readable medium. A computer-readable medium includes both computer storage media and communication media, encompassing any medium that facilitates the transfer of a computer program from one location to another. A storage medium may be any available medium accessible to a computer. By way of example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and is accessible to a computer. Any connection is also legitimately referred to as a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of a medium. As used in this article, disk and disc include compact discs (CDs), laser discs, optical discs, digital multi-purpose discs (DVDs), floppy disks, and Blu-ray discs. Disks typically reproduce data magnetically, while discs reproduce data optically using lasers. Combinations of these should also be included within the scope of computer-readable media.

Claims

1. A multidisciplinary design simulation optimization parallel computing system, characterized in that, This includes local workstations integrating multiple optimization layers and cloud-based supercomputers that communicate and interact with each optimization layer individually; among them, The local workstation deploys an optimization layer parallel strategy allocation module and an optimization parallel strategy module. After the local workstation initiates a multidisciplinary design optimization task, it allocates the corresponding parallel optimization strategy from the optimization parallel strategy module to the corresponding optimization layer based on the initiated multidisciplinary design optimization task. This allows each optimization layer to establish its corresponding optimization calculation task based on the allocated parallel optimization strategy and upload it to the cloud supercomputer for solving. The cloud-based supercomputer deploys a simulation parallel strategy module, a resource scheduling module, and a parallel algorithm library. After receiving optimization computation tasks uploaded from each optimization layer, the cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each optimization computation task through the simulation parallel strategy module and the resource scheduling module. Then, based on the allocated simulation parallel strategies and computing resources, it calls parallel algorithms from the parallel algorithm library to solve the optimization computation tasks of each optimization layer, and sends the solution results of each optimization computation task to the corresponding optimization layer for local iterative optimization until the entire multidisciplinary design optimization task is completed.

2. The multidisciplinary design simulation optimization parallel computing system according to claim 1, characterized in that, The optimization layer is divided into the whole machine layer, component layer, part layer, and discipline layer. After the local workstation initiates a multidisciplinary design optimization task, it assigns corresponding parallel optimization strategies to the integrated whole machine layer, component layer, part layer, and discipline layer. Among them, the parallel optimization strategies include parallel allocation strategies for the whole machine layer, component layer, part layer, and discipline layer. After each optimization layer obtains the assigned parallel optimization strategy, it establishes a corresponding optimization calculation task based on the assigned parallel optimization strategy, and then uploads the established optimization calculation task to the cloud supercomputer for solving.

3. The multidisciplinary design simulation optimization parallel computing system according to claim 2, characterized in that, After receiving the optimization computation tasks uploaded by each optimization layer, the cloud supercomputer allocates the corresponding simulation parallel strategy from the simulation parallel strategy module to the corresponding optimization computation task according to the subject module corresponding to each optimization computation task. Then, based on the allocated simulation parallel strategy, the corresponding optimization computation task is decomposed into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solved, thereby completing the solution of the optimization computation tasks uploaded by each optimization layer.

4. The multidisciplinary design simulation optimization parallel computing system according to claim 3, characterized in that, The subject modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies according to the subject module corresponding to each optimization calculation task. Each subject module then decomposes the corresponding optimization calculation task into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solves them based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies.

5. The multidisciplinary design simulation optimization parallel computing system according to claim 4, characterized in that, The cloud-based supercomputer decomposes the optimization tasks uploaded from each optimization layer into multiple task packages. A resource scheduling module allocates corresponding computing resources to each task package, and then solves the tasks based on these allocated resources. The resource scheduling module includes an MPI accelerator, a computational load statistical analysis program, and a load balancing resource scheduler. When allocating computing resources to each task package, the resource scheduling module first performs MPI compilation and transformation on each task package using the MPI accelerator. Then, the compiled and transformed task packages are distributed to each solution node. The computational load statistical analysis program estimates the computational load of each transformed task package. Finally, the load balancing resource scheduler allocates corresponding computing resources to each task package, thereby utilizing the allocated computing resources to solve the task packages on each solution node.

6. The multidisciplinary design simulation optimization parallel computing system according to claim 5, characterized in that, After the cloud-based supercomputer completes the resource allocation for each task package, it calls the parallel algorithms in the parallel algorithm library to solve the task packages at each solution point. Among them, the parallel algorithms include the direct method for parallel computation of linear equations, the iterative method for parallel computation of linear equations, the domain decomposition parallel computation method, and the hierarchical parallel computation method.

7. The multidisciplinary design simulation optimization parallel computing system according to claim 6, characterized in that, After the cloud-based supercomputer completes the solution of the task packages corresponding to each optimization calculation task, it sends the solution results of each task package to the optimization layer on the corresponding local computer. After receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results of each optimization layer.

8. The multidisciplinary design simulation optimization parallel computing system according to claim 1, characterized in that, When each optimization layer performs local iterative optimization, it also introduces the optimization result of the previous layer as input parameters to solve for the optimization result, and uses the optimization result of the next layer as a constraint condition to solve for the optimization result of the current layer.

9. The multidisciplinary design simulation optimization parallel computing system according to claim 1, characterized in that, The multidisciplinary design simulation optimization parallel computing system also includes a data transmission module for data transmission between optimization layers on the local computer, between optimization layers and the cloud supercomputer, and between and within solution nodes of the cloud supercomputer. The data transmission module includes: a data transmission load balancer, an inter-node transmission controller, an intra-node transmission controller, and a local / supercomputer upload / download controller.

10. The multidisciplinary design simulation optimization parallel computing system according to claim 1, characterized in that, The multidisciplinary design simulation optimization parallel computing system also includes a high-performance auxiliary module, which provides the supercomputing resources required for simulation optimization parallel computing for the cloud supercomputer; wherein, the high-performance auxiliary module includes: a supercomputing high-speed computing management unit, a supercomputing high-speed computing storage unit, a supercomputing high-speed computing unit, and a supercomputing high-speed network unit.

11. A multidisciplinary design simulation optimization parallel computing method, characterized in that, Includes the following steps: Step S1: Initiate a multidisciplinary design optimization task based on a local workstation with multiple optimization layers deployed; Step S2: Assign corresponding parallel optimization strategies to each optimization layer of the local workstation based on the initiated multidisciplinary design optimization task; Step S3: Each optimization layer establishes a corresponding optimization computation task based on the assigned parallel optimization strategy and uploads it to the cloud supercomputer for solving; Step S4: The cloud-based supercomputer allocates corresponding simulation parallel strategies and computing resources to each received optimization computing task; Step S5: The cloud-based supercomputer uses parallel algorithms from the parallel algorithm library to solve the optimization tasks of each optimization layer based on the allocated simulation parallel strategy and computing resources, and obtains the solution results of each optimization task. Step S5: The cloud-based supercomputer distributes the solution results of each optimization calculation task to the corresponding optimization layers for local iterative optimization until the entire multidisciplinary design optimization task is completed.

12. The multidisciplinary design simulation optimization parallel computing method according to claim 11, characterized in that, The optimization layer is divided into the whole machine layer, component layer, part layer, and discipline layer. After the local workstation initiates a multidisciplinary design optimization task, it assigns corresponding optimization parallel strategies to the integrated whole machine layer, component layer, part layer, and discipline layer. After each optimization layer obtains the assigned optimization parallel strategy, it establishes a corresponding optimization calculation task based on the assigned parallel optimization strategy, and then uploads the established optimization calculation task to the cloud supercomputer for solving.

13. The multidisciplinary design simulation optimization parallel computing method according to claim 12, characterized in that, After receiving the optimization calculation tasks uploaded by each optimization layer, the cloud supercomputer assigns a corresponding simulation parallel strategy to each optimization calculation task according to the subject module corresponding to each optimization calculation task. Then, based on the assigned simulation parallel strategy, the corresponding optimization calculation task is decomposed into multiple task packages that can be computed on the solution nodes of the cloud supercomputer and solved, thereby completing the solution of the optimization calculation tasks uploaded by each optimization layer.

14. The multidisciplinary design simulation optimization parallel computing method according to claim 13, characterized in that, The subject modules are divided into explicit dynamics, statics, aerodynamics, and combustion. The cloud supercomputer allocates corresponding explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies according to the subject module corresponding to each optimization calculation task. Then, based on the allocated explicit dynamics simulation parallel strategies, statics simulation parallel strategies, aerodynamics simulation parallel strategies, and combustion simulation parallel strategies, the corresponding optimization calculation tasks are decomposed into multiple task packages that can be computed on the cloud supercomputer's solution nodes and solved.

15. The multidisciplinary design simulation optimization parallel computing method according to claim 14, characterized in that, The cloud-based supercomputer decomposes the optimization computation tasks uploaded from each optimization layer into multiple task packages, allocates corresponding computing resources to each task package, and then solves each task package based on the allocated computing resources. When allocating computing resources to each task package, the multidisciplinary design simulation optimization parallel computing method first performs MPI compilation and transformation on each task package using an MPI accelerator, then distributes the compiled and transformed task packages to each solver node, estimates the computational workload of each transformed task package using a computational workload statistical analysis program, and finally allocates corresponding computing resources to each task package through a load balancing resource scheduler, thereby utilizing the allocated computing resources to solve the task packages on each solver node.

16. The multidisciplinary design simulation optimization parallel computing method according to claim 15, characterized in that, After the cloud-based supercomputer completes the resource allocation for each task package, it calls the parallel algorithms in the parallel algorithm library to solve the task packages at each solution point. Among them, the parallel algorithms include the direct method for parallel computation of linear equations, the iterative method for parallel computation of linear equations, the domain decomposition parallel computation method, and the hierarchical parallel computation method.

17. The multidisciplinary design simulation optimization parallel computing method according to claim 16, characterized in that, After the cloud-based supercomputer completes the solution of the task packages corresponding to each optimization calculation task, it sends the solution results of each task package to the optimization layer on the corresponding local computer. After receiving the solution results, each optimization layer performs local iterative optimization based on the solution results sent from the cloud as input parameters, thereby obtaining the optimization results of each optimization layer.

18. The multidisciplinary design simulation optimization parallel computing method according to claim 1, characterized in that, When each optimization layer performs local iterative optimization, it also introduces the optimization result of the previous layer as input parameters to solve for the optimization result, and uses the optimization result of the next layer as a constraint condition to solve for the optimization result of the current layer.

19. A computer-readable medium storing computer program code, characterized in that, The computer program code, when executed by a processor, implements the method as described in any one of claims 11-18.

20. A multidisciplinary design simulation optimization parallel computing device, characterized in that, include: Memory is used to store instructions that can be executed by the processor; as well as A processor for executing the instructions to implement the method as described in any one of claims 11-18.

Images