Aero-Engine Transition State Prediction Using Auto-Encoder and Random Forest
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for predicting aero-engine transition state acceleration process parameters are hindered by high requirements for model parameters and input features, poor generalization ability, and inefficiency in handling high-dimensional data, leading to increased human and financial costs due to the need for frequent re-selection of model parameters and features.
Innovation Solution
A method utilizing the Random Forest algorithm for parameter prediction, combined with a neural network-based sparse auto-encoder for feature dimensionality adjustment, optimized using a particle swarm algorithm to improve prediction accuracy and efficiency, particularly for high-dimensional data in the context of aero-engine performance parameter forecasting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional regression prediction algorithms are used, then the model can be built with simple structure, but the prediction accuracy on high-dimensional data is poor
Solution Approach 1:
The patent transforms the input feature space from original dimension to a higher-dimensional space through kernel mapping. The support vector regression algorithm implicitly maps input data into a high-dimensional feature space where nonlinear relationships become linearly separable, thereby improving prediction accuracy without significantly increasing model structural complexity.
Solution Approach 2:
The patent changes the parameter representation by introducing kernel parameters (gamma, sigma) and regularization parameter (C) that control the complexity and fit of the model. By optimizing these parameters, the model achieves better prediction accuracy on high-dimensional aero-engine data while maintaining a relatively simple structural framework.
2Productivity
If traditional machine learning algorithms are used, then the model can handle basic prediction tasks, but the generalization ability is poor when engine types change
Solution Approach 1:
The patent enhances model universality by integrating multiple algorithms (support vector regression, particle swarm optimization, and feature selection mechanisms) into a unified framework. This hybrid approach enables the model to adapt to different engine types and prediction scenarios, improving generalization ability while maintaining strong prediction capability across various conditions.
3Measurement precision
If model parameters and input features are frequently re-selected for different engines, then the prediction accuracy for each engine can be optimized, but the human and financial consumption increases
Solution Approach 1:
The patent performs preliminary feature selection and parameter optimization through the integrated framework that automatically identifies relevant features and optimizes parameters before actual prediction. This preliminary processing reduces the need for repeated parameter selection when dealing with different engine types, thereby saving time and resources while maintaining prediction accuracy.
Data Source
AI summary
A method for prediction of key performance parameters of an aero-engine transition state acceleration process based on space reconstruction. Aero-engine transition state acceleration process test data provided by a research institute is used for establishing a training dataset and a testing dataset; dimension increase is conducted on the datasets based on the data space reconstruction of an auto-encoder; model parameters optimization is conducted by population optimization algorithms which is represented by particle swarm algorithm; and random forest regression algorithm performing well on high-dimensional data is used for carrying out regression on transition state performance parameters, which realizes effective real-time prediction from the perspective of engineering application.