CN-122014423-A - Parameter optimization method and device based on distributed artificial intelligence hybrid modeling

CN122014423ACN 122014423 ACN122014423 ACN 122014423ACN-122014423-A

Abstract

The invention relates to a parameter optimization method and device based on distributed artificial intelligence hybrid modeling, wherein the method comprises the steps of obtaining a target optimization task aiming at a target engine system, determining a target adjustment direction aiming at the target engine system according to the target optimization task, inputting the target adjustment direction into an optimization model to obtain an optimization result, wherein the optimization result comprises parameter adjustment amounts of components required by meeting the target adjustment direction, the optimization model comprises a plurality of distributed artificial intelligence AI bodies, each distributed AI body corresponds to one or more components, each distributed AI body is used for optimizing parameters of the corresponding component, and a neural network layer in the distributed AI bodies corresponds to an AI model or a mechanism model. The method and the device can effectively process the part characteristics of the analytic formula and learn the complex nonlinear mapping relation, and greatly improve the efficiency and the precision of engine modeling.

Inventors

FENG XUDONG
XU QIANNAN
WU FENG
ZHANG HAI
HAN YU

Assignees

中国航发四川燃气涡轮研究院

Dates

Publication Date: 20260512
Application Date: 20260116

Claims (10)

1. A method for optimizing parameters based on distributed artificial intelligence hybrid modeling, the method comprising: Acquiring a target optimization task for a target engine system, the target engine system comprising a plurality of components; determining a target adjustment direction for a target engine system according to the target optimization task; inputting the target adjustment direction into an optimization model to obtain an optimization result, wherein the optimization result comprises parameter adjustment amounts of components required for meeting the target adjustment direction, the optimization model comprises a plurality of distributed Artificial Intelligence (AI) bodies, each AI body corresponds to one or more components, each AI body is used for optimizing parameters of the corresponding component, and a neural network layer in the AI body corresponds to an AI model or a mechanism model.
2. The method of claim 1, further comprising at least one of: Corresponding all parts moving at the same rotating speed in the target engine system to the same distributed AI body; corresponding all parts in the same energy conversion path in the target engine system to the same distributed AI body; Corresponding all parts in the same fault propagation range in the target engine system to the same distributed AI body; and corresponding all the components synchronously controlled in the target engine system to the same distributed AI body.
3. The method according to claim 1, wherein the method further comprises: And constructing a corresponding working condition neural network aiming at each working state of the target engine system, wherein each working condition neural network comprises a plurality of component layers, each component layer corresponds to different components, and the optimization model comprises the working condition neural network of each working state.
4. A method according to any one of claims 1 to 3, further comprising: Determining each training sample according to historical operation data of each component of a target engine system to form a training sample set, wherein each training sample comprises operation parameters externally input to the component and test result data of the operation of the component; And training the initial optimization model by using the training sample set until the trained optimization model meets the preset condition, and outputting the trained optimization model.
5. The method of claim 4, wherein training the optimization model using the training sample set until the trained optimization model meets a preset condition, outputting a trained optimization model, comprising: In each working state, determining a maximum value item under the corresponding working state according to the maximum value output by each component layer in the corresponding working condition neural network, determining output values of all component layers in the corresponding working condition neural network, and determining a weighted sum item under the corresponding working state according to all the output values and a preset dynamic balance coefficient; Determining first losses under the corresponding working states according to the maximum value item and the weighted sum item under each working state, and taking the sum of the first losses under all working states as a loss function value used in the training of the optimization model; and outputting the current trained optimization model under the condition that the loss function value is smaller than a preset threshold value.
6. The method of claim 4, wherein the predetermined condition further comprises a number of iterations meeting a predetermined iteration threshold.
7. A distributed artificial intelligence hybrid modeling-based parameter optimization device, the device comprising: An acquisition module for acquiring a target optimization task for a target engine system, the target engine system comprising a plurality of components; the determining module is used for determining a target adjusting direction aiming at a target engine system according to the target optimizing task; The optimization module is used for inputting the target adjustment direction into an optimization model to obtain an optimization result, wherein the optimization result comprises parameter adjustment amounts of components required for meeting the target adjustment direction, the optimization model comprises a plurality of distributed Artificial Intelligence (AI) bodies, each AI body corresponds to one or more components, each AI body is used for optimizing parameters of the corresponding component, and a neural network layer in the AI bodies corresponds to an AI model or a mechanism model.
8. A distributed artificial intelligence hybrid modeling based parameter optimization apparatus comprising a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the method of any one of claims 1 to 6.
9. A non-transitory computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the method according to any of claims 1 to 6.
10. A computer program product comprising a computer program, or a non-transitory computer readable storage medium carrying a computer program, characterized in that the computer program when executed by a processor implements the steps of the method according to any one of claims 1 to 6.

Description

Parameter optimization method and device based on distributed artificial intelligence hybrid modeling Technical Field The disclosure relates to the technical field of engine optimization, in particular to a parameter optimization method and device based on distributed artificial intelligence hybrid modeling. Background In an aeroengine system, how to construct a distributed artificial intelligence (ARTIFICIAL INTELLIGENCE, AI) architecture to process the complex input-output relationship of a component layer and the coupling optimization problem of a prediction and a system layer, specifically, the method comprises the steps of utilizing a nonlinear input-output mapping relationship which cannot be accurately described by a traditional formula in key components of an AI intelligent agent learning part, simultaneously establishing a parameter optimization algorithm capable of processing the complex coupling influence of a preamble component on a follow-up component, realizing global parameter optimization under various working conditions, and ensuring that a set of optimal parameter combination capable of adapting to various working conditions is found, thereby solving the technical bottleneck faced by the traditional engine modeling method in processing the nonlinear coupling relationship among the components and optimizing the parameters under the multiple working conditions. Disclosure of Invention In view of this, the present disclosure proposes a parameter optimization method and apparatus based on distributed artificial intelligence hybrid modeling. According to one aspect of the disclosure, a parameter optimization method based on distributed artificial intelligence hybrid modeling is provided, and the method comprises the steps of obtaining a target optimization task for a target engine system, wherein the target engine system comprises a plurality of components, determining a target adjustment direction for the target engine system according to the target optimization task, and inputting the target adjustment direction into an optimization model to obtain an optimization result, wherein the optimization result comprises parameter adjustment amounts of the components required for meeting the target adjustment direction, the optimization model comprises a plurality of distributed artificial intelligence AI bodies, each distributed AI body corresponds to one or more components, each distributed AI body is used for optimizing parameters of the corresponding components, and a neural network layer in the distributed AI bodies corresponds to an AI model or a mechanism model. The distributed artificial intelligent hybrid modeling parameter optimization method is characterized in that a neural network layer and a mechanism model are deeply fused, an optimization model has both data fitting capacity and physical interpretability, distributed parallel computing can shorten the optimization period of complex working conditions, and rapid mapping of a target adjustment direction is realized through a reinforcement learning training strategy aiming at dynamic change of a target optimization task. In one possible implementation, the method further comprises at least one of associating components of the target engine system moving at the same rotational speed with the same distributed AI body, associating components of the target engine system in the same energy conversion path with the same distributed AI body, associating components of the target engine system in the same fault propagation range with the same distributed AI body, and associating components of the target engine system synchronously controlled with the same distributed AI body. Thus, through the four dividing modes, the physical relevance and the control logic independence can be simultaneously met, the coupling relation of the components on the mechanical structure is reflected, and the AI is ensured to independently finish the perception-decision-execution closed-loop control. In one possible implementation, the method further includes constructing a corresponding working condition neural network for each working state of the target engine system, each working condition neural network including a plurality of component layers, each component layer corresponding to a different component, and the optimization model including the working condition neural network for each working state. The design not only completely maintains the physical connection relation and the energy transfer characteristic among the components of the engine, but also lays a solid mathematical foundation for unified expression and collaborative optimization of heterogeneous models. And each working condition neural network realizes accurate modeling of a physical rule under a specific working state through parameter coupling of component layers. The working condition neural network shares the feature extraction layer through transfer learning, and new working condition data re