Compressor Aerodynamic Noise Prediction for Fine Frequency Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for acquiring aerodynamic noise data in compressors face limitations due to a limited time step and high computing resource consumption, particularly in numerical calculations, which affect the resolution of frequency-domain noise data and require significant manpower and material resources.
Innovation Solution
A method involving inverse Fourier transform, nonlinear autoregressive neural network modeling, and Fourier transform is used to predict aerodynamic noise data, allowing for finer frequency resolution and reduced resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If numerical calculation is used to acquire aerodynamic noise data, then computing resource consumption increases and time step decreases, but frequency resolution of noise data deteriorates
Solution Approach 1:
The patent uses neural network models to copy and predict aerodynamic noise characteristics from training data, replacing the need for computationally intensive numerical calculations. The trained model can rapidly generate noise predictions without requiring complex CFD simulations, thus reducing computing resource consumption while maintaining frequency resolution.
Solution Approach 2:
The patent substitutes the mechanical numerical calculation system with an AI-based prediction system. Instead of performing complex numerical simulations that require small time steps and high computing power, the system uses trained neural networks to directly predict noise characteristics, eliminating the trade-off between computing resources and frequency resolution.
2Productivity
If numerical calculation with small time step is used, then aerodynamic noise data can be acquired, but acquisition speed decreases and computing resource consumption increases
Solution Approach 1:
The patent performs preliminary training of neural network models using training data from numerical calculations or experiments. Once trained, the model can rapidly predict aerodynamic noise characteristics without requiring real-time numerical simulations, thus achieving fast acquisition speed without the computing resource consumption associated with small time step calculations.
Solution Approach 2:
The trained neural network model copies the relationships between input parameters and noise characteristics learned from training data, enabling rapid prediction without repeating the computationally intensive numerical calculation process. This copying approach maintains accuracy while dramatically improving acquisition speed.
3Measurement precision
If experimental testing is used to acquire aerodynamic noise data, then data can be obtained rapidly and accurately, but personnel and equipment requirements increase
Solution Approach 1:
The patent uses neural network models to copy aerodynamic noise characteristics from training data, eliminating the need for physical experimental testing equipment and personnel. The model can generate accurate noise predictions by learning from existing data, thus maintaining measurement precision while reducing equipment and personnel requirements to minimal computational resources.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables rapid and accurate acquisition of finer aerodynamic noise data, increasing frequency resolution while minimizing computing and material resources, facilitating more detailed noise analysis and optimization of compressor design.
Implementation Method 1
performing inverse Fourier transform on the first aerodynamic noise frequency-domain data to obtain corresponding first aerodynamic noise time-domain data
Implementation Method 2
outputting a sound pressure corresponding to each future time step using a trained nonlinear autoregressive neural network model
Implementation Method 3
performing Fourier transform on the second aerodynamic noise time-domain data to obtain second aerodynamic noise frequency-domain data
Data Source
AI summary
A method and apparatus for acquiring aerodynamic noise of a compressor, a medium, and a product are provided. The method includes: performing inverse Fourier transform on a known frequency-domain noise spectrum to obtain known time-domain noise data, then predicting unknown time-domain noise using a time series neural network, obtaining finer time-frequency noise data in combination with the known time-frequency noise data, and finally, performing Fourier transform on the finer time-frequency noise data to obtain new frequency-domain noise data. Limitations of a limited time step and a total simulation time on the acquisition of aerodynamic noise data in traditional numerical calculation of aerodynamic noise can be overcome. Finer aerodynamic noise data can be acquired rapidly and accurately. The frequency resolution of the noise spectrum can be increased, and reducing the consumption of computing resources and saving manpower and material resources can be achieved.