A method for predicting engine exhaust nozzle wall pressure by integrating experimental data
By combining experimental data and CFD data using a combined neural network, the problem of obtaining the pressure on the engine exhaust nozzle wall was solved, enabling rapid and high-precision prediction, shortening the design cycle and reducing costs.
CN116644510BActive Publication Date: 2026-05-26NORTHWESTERN POLYTECHNICAL UNIV
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NORTHWESTERN POLYTECHNICAL UNIV
- Filing Date
- 2023-05-07
- Publication Date
- 2026-05-26
AI Technical Summary
Technical Problem
In existing technologies, it is difficult to obtain the pressure on the engine exhaust nozzle wall, the CFD calculation results differ greatly from the experimental measurement results, the design cycle is long and the cost is high, and the experimental setup is difficult.
Method used
A combined neural network model was constructed, and experimental data and CFD data were fused using fully connected layers and convolutional neural networks. A nozzle wall pressure prediction method was established through iterative optimization to quickly obtain high-precision data.
Benefits of technology
It enables rapid and accurate prediction of nozzle wall pressure, reduces experimental costs and time, improves design efficiency, and overcomes the errors of CFD calculations.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116644510B_ABST
Abstract
This invention discloses a method for predicting engine exhaust nozzle wall pressure by fusing experimental data. First, high-fidelity data of nozzle wall pressure measured experimentally under different operating conditions are discretized. A model is constructed using a combined neural network. For different operating conditions of the nozzle, aerodynamic / geometric feature data of the nozzle is input. First, a fully connected layer is used to establish a mapping relationship between aerodynamic / geometric parameters and a small amount of experimental pressure data. Then, the experimental data is fused with CFD low-fidelity data. A convolutional neural network is used to perform convolution operations on the data to extract feature data and output the nozzle wall pressure. The method is iteratively optimized until the loss requirement is met. The model training optimization objective is set to the experimental high-fidelity data of the nozzle under different operating conditions. The engine exhaust nozzle wall pressure prediction method using this invention can quickly predict nozzle wall pressure while ensuring a certain level of prediction accuracy and generalization.
Need to check novelty before this filing date? Find Prior Art